CATT Manual v8

CATT-Acoustic v8

g

Room Acoustics Prediction and Walkthrough Auralization

User’s Manual

Copyright © CATT 2007

CATT-Acoustic v8.0 - up to and including v8.0g build 2 For additions after v8.0g build 2 see Help|View Manual Addendum or the help-file

User’s Manual Copyright © CATT 2007

CATT Mariagatan 16A SE-41471 Gothenburg Sweden Phone/Fax: +46 31 145154 E-mail: [email protected] WWW: http://www.catt.se

Front page pictures and model from Javier Barros Sierra Auditorium at Facultad de Ingenieria, UNAM, Mexico Courtesy of Juan Carlos Andrews Gonzalez from his thesis on the acoustic analysis of this hall. Printed in Gothenburg Sweden, Aug 2007

CONTENTS

1. INTRODUCTION .......................................................................1 1.1 Installation............................................................................................ 1 Optional user-supplied info included in license..........................................................6

1.2 Typical session.................................................................................... 6 Audience area mapping ...............................................................................................10 Early part detailed ISM .................................................................................................10 Full detailed calculation ...............................................................................................10

1.3 Common user interface elements .................................................... 11 Menus.............................................................................................................................11 Dialogs...........................................................................................................................19 Toolbars.........................................................................................................................19 Hidden options..............................................................................................................20 Miscellaneous ...............................................................................................................23

1.4 Input/output files overview ............................................................... 25 Source directivity module............................................................................................26 Surface properties module ..........................................................................................26 Prediction module ........................................................................................................26 Source addition module...............................................................................................27 Post-processing module..............................................................................................28 Plot-file viewer module.................................................................................................29 Sequence processing module.....................................................................................29

1.5 Software overview ............................................................................. 29 Hardware requirements................................................................................................29 Hall geometry and absorption .....................................................................................30 Coordinate system .......................................................................................................31 Frequency range...........................................................................................................32 Source directivity..........................................................................................................32 Data output....................................................................................................................32

1.6 How to get started ............................................................................. 34

CONTENTS

1.7 About CATT........................................................................................ 34

2. PREDICTION MODULE.......................................................... 37 2.1 Menus ................................................................................................. 37 2.2 Dialogs................................................................................................ 43 General settings ........................................................................................................... 44 Geometry view/check .................................................................................................. 48 Audience area mapping............................................................................................... 53 Early part detailed ISM................................................................................................. 60 Full detailed calculation .............................................................................................. 63 Interactive RT estimate................................................................................................ 69 Pixel rendering ............................................................................................................. 72

2.3 Input-files ........................................................................................... 76 Common features......................................................................................................... 76 Geometry-files .............................................................................................................. 78 Source-file................................................................................................................... 110 Receiver-file ................................................................................................................ 115

2.4 Output-files ...................................................................................... 121 General ........................................................................................................................ 121 Geometry view/check results.................................................................................... 121 Audience area mapping results................................................................................ 123 Early part detailed ISM results .................................................................................. 125 Full detailed calculation results................................................................................ 127

2.5 Theory............................................................................................... 132 Prediction methods.................................................................................................... 132 Automatic edge diffusion .......................................................................................... 136 Reverberant decay and reverberation time estimation .......................................... 137 Room acoustic parameters ....................................................................................... 138 Echogram calculation ................................................................................................ 143 Directivity handling .................................................................................................... 144 Semi-transparency ..................................................................................................... 145

CONTENTS Calculation recommendations ..................................................................................146

2.6 Tips and Tricks ................................................................................ 147 2.7 Trouble-shooting ............................................................................. 150 Geometrical errors......................................................................................................150 Audience area mapping problems............................................................................151 Early part calculation problems ................................................................................151 Full detailed calculation problems............................................................................151

3. SURFACE PROPERTIES MODULE .....................................153 3.1 Menus ............................................................................................... 153 3.2 Dialogs.............................................................................................. 157

4. DIRECTIVITY MODULE ........................................................159 4.1 Menus ............................................................................................... 159 4.2 Dialogs.............................................................................................. 170 15-degree interpolated format (SD0) ........................................................................170 10-degree measured format (SD1, CF1, CF2) ..........................................................171 DLL-format (SD2) ........................................................................................................172 Elements common to SD0, SD1, CF1, CF2 (and SD2) formats...............................173

4.3 Theory............................................................................................... 176 The 15-degree interpolated format (SD0) .................................................................176 The 10-degree measured format (SD1).....................................................................176 The any-degree measured, interpolated or simulated format (SD2) .....................177 Directivity Index ..........................................................................................................178 Directivity Contours ...................................................................................................178 Caching of DI and contours.......................................................................................178

5. PLOT-FILE VIEW/PRINT MODULE ......................................179 5.1 Menus ............................................................................................... 179 5.2 Dialogs.............................................................................................. 184 Plot-file control ...........................................................................................................184

6. SOURCE ADDITION MODULE .............................................187 6.1 Menus ............................................................................................... 188

CONTENTS

6.2 Dialogs.............................................................................................. 189 6.3 Output-files ...................................................................................... 192 General ........................................................................................................................ 192 Single receiver result-files......................................................................................... 192 Multiple receiver result-files...................................................................................... 194

7. POST-PROCESSING MODULE ........................................... 197 7.1 Menus ............................................................................................... 198 7.2 Dialogs.............................................................................................. 224 General settings ......................................................................................................... 225 Output-files ................................................................................................................. 229 Early part response.................................................................................................... 231 Late part response ..................................................................................................... 232 Create files for head-tracking ................................................................................... 232 Create CWI-file for CATT-Walker .............................................................................. 233 Convolution ................................................................................................................ 234 Walkthrough convolution.......................................................................................... 236 Convolution (WAV impulse responses)................................................................... 240

7.3 Theory............................................................................................... 242 7.3 Theory............................................................................................... 242 Impulse response scale and calibration .................................................................. 242 Reflection path (RP) transfer function ..................................................................... 243 Receiver model........................................................................................................... 245 Late part generation................................................................................................... 247 Inter-Aural Cross Correlation Coefficient ................................................................ 248 Crosstalk canceling filters ........................................................................................ 248

8. SEQUENCE PROCESSING MODULE ................................. 253 9. CATT EDITOR ...................................................................... 263 9.1 Non-standard menus....................................................................... 263

10. AUTOCAD INTERFACE ..................................................... 265 10.1 Introduction.................................................................................... 265

CONTENTS Direct creation of a model in AutoCAD ....................................................................265 Modification of an existing drawing in AutoCAD format ........................................266 Import of an existing model in the CATT format .....................................................266 Basic interface concepts ...........................................................................................268

10.2 Installation...................................................................................... 268 10.3 AutoLISP commands..................................................................... 268 10.4 Modeling tips.................................................................................. 271

11. TUTORIAL...........................................................................273 11.1 Preparation..................................................................................... 273 11.2 Modeling ......................................................................................... 273 Step 1, the audience and floor surfaces...................................................................275 Step 2, walls and ceiling ............................................................................................277 Step 3, entrance wall and door..................................................................................278 Step 4, the stage .........................................................................................................279 Step 5, source/receiver locations..............................................................................280 Step 6, side wall reflectors.........................................................................................282

11.3 Prediction ....................................................................................... 283 Interactive RT estimate ..............................................................................................283 Audience area mapping .............................................................................................284 Early part detailed ISM ...............................................................................................284 Full detailed calculation .............................................................................................285

11.4 Source addition.............................................................................. 286 11.5 Post-processing - Auralization ..................................................... 288 11.6 Sequence processing.................................................................... 289

12. REFERENCES ....................................................................291 APPENDIX A – TUTORIAL FILES INDEX

CONTENTS

1. INTRODUCTION

1. INTRODUCTION The Introduction describes how CATT-Acoustic is installed, an example of how it is run, common user interface elements, an overview of input/output-files used, a general overview of what the software can do, and a CATT background. Manual conventions: File-names, folder-names, text to be entered and text-file contents are set in COURIER like in C:\CATT. Menu selections are set in bold as in File|Preferences. Dialog names are set in bold italics as in General settings. Dialog items are set in italics as in General settings/Input folder. Dialog groups are set in italics within brackets as in General settings/(Plane use) . Menus and dialogs belonging to specific programs or program modules are sometimes written prefixed with the module name and a colon as in Prediction:General settings. Product, organization and company names are set in italics as in CATT-Acoustic. Items in square brackets as in [Dalenbäck-95] refer to References, Section 12. All products mentioned in this manual are trademarks of their respective owners.

1.1 Installation CATT-Acoustic is installed from CD by inserting the CD (in say G:) and Run G:\install.exe. The installation program will then perform all necessary steps (for licenses including hardware locks, and when running under Windows NT, 2000, XP and Vista some further short separately described steps are required after installation, see README.TXT). All CATT programs are normally kept in a single folder selected at installation, this folder will hereafter be called the "CATT-folder". All CATT data-libraries, settings-files and the DLL Directivity Interface are kept in a separate folder. This folder will hereafter be called the "CATTDATA-folder" and "CATTDATA-folder ANECHOIC" means the ANECHOIC sub-folder. This folder is at initial installation suggested to be: XP:

C:\Documents and Settings\[user]\Application Data\CATT *)

Vista:

C:\Users\[user]\AppData\Roaming\CATT *)

If more than one logged-on user will be using the software, and want to share the library data, it can be more practical to install in XP:

C:\Documents and Settings\All Users\Application Data\CATT *)

Vista:

C:\ProgramData\CATT *)

*) or corresponding name on a non-English Windows

1

1. INTRODUCTION On older installations updated to v8.0g or higher the CATT-folder and the CATTDATA-folder may remain the same but if necessary (due to folder access rights) or desired, the data can be copied to another folder or even another drive (for that to work the drive must be mapped so it gets a drive letter). The items to copy are: Folders (including all sub-folders) for all versions: [Abslibs] [Anechoic] [HRTFs] [Headphones] [SD] [SD2Data] If exists: hiddenoptions.txt Template.PRD Template.PST Startup.PLT Startup.OGL CATT32.DAT For full versions with The FIReverb Suite : [pv_mics] [pv_halls] [mv_mics] FIReverb.DAT For full versions with CATT-Walker: BinDec.DAT Walker.CFG Multiple data-folders can exist and the one to use can be selected by File|Preferences. No files are installed in the WINDOWS folder and no system files other than the Registry are altered. The Registry is used for Preferences settings and file types/icons. After installation, the files and folders below are present in the CATT-folder and CATTDATAfolder. Files marked "internal" may only be changed, removed or updated after instructions from CATT or by CATT software. CATT programs (internal, (f) = full auralization version), some of these can be loaded from within CATT-Acoustic via File|CATT External Tool: CATT-Acoustic main program CATT-Edit (Notepad-like customized editor) CATT 3D-viewer based on OpenGL (documented by its own help-file) CATTPLT32.EXE CATT PLT-viewer (standalone version) DXF2GEO.EXE DXF2GEO DXF converter (documented only by its own help PDF) MAKEWALKERBINDEC.EXE (f) MakeWalkerBinDec support tool for CATT-Walker (see CATTWalker help PDF) UNCTT.EXE program used for extracting software updates and additions downloaded over the internet from the CATT users’ www. CATT32.EXE CATTED32.EXE CATTOGL32.EXE

2

1. INTRODUCTION WALKDECIM.EXE (f) WALKER.EXE (f) PureVerb32.EXE (f) MltVlv32.EXE (f) PvxViewer.EXE (f) CATTxxxx_data.EXE

WalkDecim, support tool for making use of exported camera paths with the Walkthrough convolver (documented by its own help PDF). CATT-Walker for real time walkthrough auralization (documented by its own help PDF) The FIReverb Suite PureVerb application The FIReverb Suite MultiVolver application The FIReverb Suite PvxViewer application An installation program that installs the library data and establishes the CATTDATA folder when the program is first run or when run by a new user. xxxx is either demo, pred or full depending on version.

Supporting Dynamic Link Libraries (internal): ACLIB.DLL (f) DATA32.DLL ERRORS32.DLL GENLOCK.DLL METER32.DLL OGLSWITCH32.DLL STRNGS32.DLL SWITCH32.DLL License-file: XTRA32.DAT

(licenses before v8.0g use an XTRA32.DLL)

Windows Help (.HLP), HTML Help (.CHM) or Adobe Acrobat (.PDF) (internal): programname.CHM programname.HLP programname.PDF addendum.HTM/.PDF

Compiled HTML help-file Windows help-file PDF help Manual addendum (additions since this manual)

Surface properties (absorption and scattering coefficients) library files to be selected in Preferences: CATTDATA-folder ABSLIBS (folder, internal) *.DAT Source directivity library-files: CATTDATA-folder SD or SD sub-folder (folder, internal): *.SD0 (15° format ) *.SD1 (10° format) *.SD2 (DDI format) *.CF1 (Common Loudspeaker Format [CLF] 10°) *.CF2 (Common Loudspeaker Format [CLF] 5°) *.COL (directivity collection files) *.MRK (loudspeaker visuals) DLL Directivity Interface (DDI) folders used by the SD2 directivity format:

3

1. INTRODUCTION

CATTDATA-folder SD2DATA (folder, internal) CATTDATA-folder SD2DATA\CATT_Generic (folder, internal): CATT_Generic.DLL CATT_Generic.HLP CATTDATA-folder SD2DATA\… further DDI folders/modules… Headphone equalization filters to be selected in Preferences : CATTDATA-folder HEADPHONES (folder, internal): *.DAT Note: For limited versions only one headphone filter is provided: "Generic" (filename GENERIC_v8_plain_44.MAT. To use the correct headphone filter is very important. The generic filter may result in severe coloration with some headphones. Head-Related Transfer Functions (HRTFs) for binaural post-processing to be selected in Preferences: CATTDATA-folder HRTFS (folder, internal): *.DAT Note: for limited versions only an HRTF library based on analytical sphere expressions is provided: SPHERE_plain_44.DAT. Mono WAV-files with anechoically recorded material to be used with the software convolver in the Post-processing module. Further samples can be taken from the accompanying CD (full version): CATTDATA-folder ANECHOIC (folder): ????_44_AN.WAV a number of anechoic 44100 Hz WAV-files Note: these files are short samples from the accompanying CD from Bang & Olufsen (B&O). CATT has, as a service to users, been granted permission to include short samples from this CD provided that the full CD also is included with the software. Copyright remains with B&O and must be carefully considered if samples from the CD are to be used in any context covered by copyright laws such as e.g. broadcasting or internet presentation. For limited versions only a single - a bit noisy - 22050 Hz file is provided: DEMO_22_AN.WAV. *.AWI WAVLIST.TXT

anechoic WAV-file info files that contain level and directivity information associated with an anechoic WAV-file. a list and a short description of the anechoically recorded samples provided (full version).

Simple shoebox used for typical session in Section 1.2: CATTDATA-folder MODELS\SHOEBOX (folder): MASTER.GEO SRC.LOC, REC.LOC

4

1. INTRODUCTION GEO.PRD, PROJECT.TXT Manual example hall used in Section 2.3: CATTDATA-folder MODELS\EXAMPLE (folder): MASTER.GEO, SIDES.GEO, ACROSS.GEO, AUDIENCE.GEO, REFL.GEO, MARKERS.GEO SRC.LOC, REC.LOC GEO.PRD, PROJECT.TXT Tutorial hall files used in Section 11: CATTDATA-folder MODELS\TUTORIAL (folder): MASTSTEP.GEO, MASTFULL.GEO, REFLECT.GEO SRC.LOC, REC.LOC GEO.PRD, PROJECT.TXT Marker files (illustrating the use of markers for non-acoustical visual elements and loudspeaker visuals): CATTDATA-folder MARKERS (folder): *.GEO *.MRK

(NOROOM GEO-files to be exported as markers) (sample marker-file exported from corresponding GEO-files)

AutoCAD interface files (internal): CATT-folder ACAD (folder) Various AutoLISP files (*.LSP) and predefined symbols (*.DWG). The CATT AutoLISP AutoCAD interface is documented in Section 10. For full auralization licenses, The FIReverb Suite is included containing the FIR reverberation creator PureVerb and the 8x8 multi-channel convolver MultiVolver. The FIReverb Suite comes with its own help-files but no printed documentation. The FIReverb Suite consists of the following main files in the CATT-folder (for further details consult the FIReverb help-files): PUREVERB32.EXE (f) MLTVLV32.EXE (f)

PureVerb (+ help-file PUREVERB32.CHM) MultiVolver (+ help-file MLTVLV32.CHM)

Other files in the CATT-folder: *.MANIFEST

these files are included for each CATT application are included that makes them use the XP interface (color scheme, mouse-over highlight etc.) assuming the classical theme has not been selected in Control panel. If this look is not desired, remove or rename the manifest files.

README.TXT

Important information about e.g. updates.

SAMPLE_PATH.TXT (f) Sample camera path for WalkDecim. Other files in the CATTDATA-folder: HIDDENOPTIONS.TXT Hidden options settings (see Section 1-3).

5

1. INTRODUCTION WALKER.CFG (f) BINDEC.DAT (f)

Settings for CATT-Walker. HRTF and headphone settings for CATT-Walker.

Optional user-supplied info included in license The license-file mechanism can handle optional user-supplied information. A license can be tagged so that the information must be included and displayed (if it is tampered with, the program will not run). The information is displayed in a "splash" dialog when the program starts that has to be OK-ed by the user. Similarly, each time Help|About is selected the special dialog is displayed together with the normal About dialog. This addition was triggered by one case where an acoustical consultant sponsored a CATT-Acoustic license to a university and wished for the donation to be clearly visible. This information can also be requested by other users, it involves only a replacement of the license DAT.

1.2 Typical session Below are shown typical steps predicting the acoustics of a simple shoebox-shaped hall with one source and one receiver to give you a feeling of what it is all about. This example is a "mini tutorial" and is kept as clean as possible avoiding many of the more advanced features that are used in the example hall in Section 2.3 and in the dedicated Tutorial in Section 11. This simple examples also only specifies values for 125 Hz to 4k Hz (8k and 16k Hz values are extrapolated). Note: the procedure here uses direct GEO modeling but import via an AutoCAD interface, via DXF or a modeling plugin is also possible, see Section 10. However, even when most modeling is done via file import the GEO has many options that can be beneficial to use after import so at least a basic understanding of the GEO is necessary. Start CATT-Acoustic, the Prediction module and the Plot-file viewer module opens up directly. Create a new project by selecting Utilities|Create Project. The appearing dialog expects you to enter a project name and select a folder for the hall you are going to work with (say C:\SHOEBOX). The Browse folder button in the dialog also allows you to create a new folder. When OK is clicked all basic files are created automatically, ready to be filled in: MASTER.GEO REC.LOC SRC.LOC GEO.PRD PROJECT.TXT

master geometry-file template receiver positions (a generic position inserted) source positions and data (a generic source inserted) prediction settings-file prepared for geometry check a text-file for notes about the project (optional use)

Select (Edit) Geo-file and MASTER.GEO and CATT-Edit loads with the file. Enter the geometry data as shown below (the files can also be found in the CATTDATA-folder MODELS\SHOEBOX). Data items in the geometry-file are constant and absorption declarations, corner ids and coordinates (x, y, z) and, finally, plane definitions. Comments are introduced by a semicolon ";" and blank lines are allowed. ;MASTER.GEO ;constant declarations LOCAL h = 8 ;hall height LOCAL w = 10 ;hall width GLOBAL d = 24 ;hall depth ;absorption and scattering coeff. 125Hz to 4kHz [%], RGB-color ABS audience = <40 50 60 70 80 80> L <30 40 50 60 70 80> {255

6

0 0}

1. INTRODUCTION ABS wood = <15 13 10 9 8 7> L <30 30 30 30 30 30> {255 255 0} ;- if 8k and 16k values are known they can be given after ; a colon as in <15 13 10 9 8 7 : 7 8 > otherwise they ; are extrapolated from 2k and 4k values. ;- color is optional, one will be auto generated if not given. ;- scattering coefficients are optional (but strongly recommended) ; if not given the default values in General settings will be used CORNERS ;floor corners 1 -w/2 0 2 -w/2 d 3 w/2 d 4 w/2 0 ;ceiling corners 11 -w/2 0 12 -w/2 d 13 w/2 d 14 w/2 0

0 0 0 0 h h h h

PLANES [1 floor / 4 3 2 1 / audience ] [2 ceiling / 11 12 13 14 / wood ] [3 stage wall / 1 11 14 4 / wood ] [4 rear wall / 3 13 12 2 / wood ] [5 left wall / 2 12 11 1 / wood ] [6 right wall / 4 14 13 3 / wood ]

The absorption/scattering coefficient data can either be entered directly in a geometry-file, as above, or already be defined in a library maintained by the Surface properties module, see Section 3. Save MASTER.GEO. Select (Edit) Receiver-file and REC.LOC and enter the receiver data. Data items in the receiver-file are the receiver ids (00..99) and the corresponding positions (x, y, z). ;REC.LOC RECEIVERS 1 -3

d-4

1.3

Note that global constants defined in GEO-files, in this case d, can also be used in sourceand receiver-files. Save REC.LOC. Select (Edit) Source-file and SRC.LOC and enter the source data. Data items in the source-file are source ids (A0..A9, ... , to Z0..Z9), source positions (x, y, z), source directivities, source aim points (x, y, z) (or an aim() function specifying angles) and finally sound pressure levels for the six (or eight) octave bands at 1 meter distance from the source on the source axis. Optionally a source roll angle can be added, see Section 2.3. ;SRC.LOC LOCAL src_z = 1.7 SOURCEDEFS A0 1.0 1.7 src_z OMNI 1.0 3.7 Lp1m_a = <70 73 76 79 82 85>

src_z

7

1. INTRODUCTION ;an electrocacoustic source, using the aim() function A1 2.0 1.7 src_z CATT.SD1 aim(-20,10) Lp1m_a = <70 73 76 79 82 95> ;the acoustic input to the system Gain_a = < 5 10 20 20 20 20> ;how much to amplify Lp1m_a, in dB Delay_e = 0 ;delay in ms ;same type of source as A1 but uses some shortcuts and a different aim A2 -1.0 1.7 src_z CATT.SD1 1.0 3.7 src_z Lp1m_a = lp_white 79 ;the acoustic input to the system, ;white spectrum with 79 dB at 1k Gain_a = < 5 10 20 20 20 20> ;how much to amplify Lp1m_a, in dB Delay_e = 0

Source directivities are managed by the Directivity module, see Section 4. Note how a local constant, src_z, is used to ensure that the source is aimed in the horizontal plane (aim has no practical consequence for the OMNI source used for A0). Save SRC.LOC. The Create project utility also created a basic settings-file GEO.PRD loaded when the project was created. Select General settings and browse through the various settings and then do the same with Geometry view/check. View the created geometry by clicking Save and Run. This will create basic geometry plotfiles. The current settings-file is also saved so the next time you run it will contain the same settings. Settings-files can be used later on for sequence (batch) processing, see Section 8. If an input-file has been edited and changed without having been saved afterwards, the editor containing the file is moved to the front and a prompt is displayed asking if the file should be saved before continuing. The Prediction module reads the created geometry-file, source- and receiver-files in the project folder (i.e. the Input folder in the General settings dialog) and creates files in the Output folder.

Z

Z

Y

X

5m Volume:1921m³

01 Y A0

X

Fig. 1.1 Sample shoebox plot-file VIEW4.PLT One created file, named PLT_GEO.TXT contains all created plot-file names and the View module automatically displays all files created (such plot-file lists can also be customized using the View module or a text editor to show an arbitrary assembly of results for presentations). The VIEW4.PLT file is shown in Fig. 1.1.

8

1. INTRODUCTION Next/Previous file can either be selected on the optional Plot-file control floating dialog (Ctrl+R), by pressing the PgDn/PgUp keys, or via the toolbar double-arrow buttons. The View module is described in Section 5. Pay special attention to the PLINFO.PLT plot-file that contains plot sub-frames. Next/ Previous frame can be selected on the Plot-file control where all planes can be stepped through and should have their front (reflecting side) colored with the assigned or autogenerated color and the back (non-reflecting) side with black (you can also use right and left arrow keys if the dialog is closed or inactive, the toolbar single-arrow buttons, click-drag the mouse on the yellow title bar or Options|Select Frame or using a scroll wheel on the mouse, or by direct mouse selection – toggled by the F key see status bar). If the color is not on the reflecting side, the plane corner entering order in the GEO-file has to be reversed. This can either be done by reversing the sequence of corner ids or simply by changing / to \ in the plane definition. The COLORED.PLT file shows a colored version of the hall where only planes viewed from their reflecting side are colored. By rotating the hall (click and drag the left mouse button) it is easy to detect if any planes are forgotten or have been entered reversed (e.g. a reversed ceiling will be shaded on the outside). Similarly, the SHADED.PLT file shows a shaded version of the hall where only planes viewed from their reflecting side are shaded (in an angle dependent manner). A file SHADED.OGL for the standalone OpenGL-based CATT 3D-Viewer was also created and loaded into the viewer. This viewer has many options and can be e.g. used to create presentations. The viewer is documented separately in its help-file. Select the Prediction module again (by clicking on its title or on the toolbar Prediction icon). If the geometry needs to be adjusted edit the geometry-file MASTER.GEO again by (Edit) Geofile (or simply Alt-TAB to the editor if it is still open) and repeat Save and Run. For a bit of experimenting change the line LOCAL h

= 8

;hall height

to GETLOCAL h

= 8

;hall height

and you will be able to enter the height at a screen prompt while the file is being read after Save and Run has been clicked. Let us assume that the geometry now looks OK. It is then time for some predictions of the acoustics in the hall. The software offers three different main kinds of predictions: Audience area mapping:

color mapping of parameters and/or the direct sound over selected audience planes – based on ray-tracing;

Early part detailed ISM:

detailed early reflections for qualitative reflection path studies – based on the Image Source Model (ISM);

Full detailed calculation:

full detailed echograms that can be used for auralization and parameter estimation - based on the unique Randomized Tailcorrected Cone-tracing (RTC).

9

1. INTRODUCTION

Audience area mapping Select (Audience area mapping) Direct + reflected sound and in the Audience area mapping - direct+reflected dialog select No of rays/octave 10000, Ray truncation time 1000 ms. Click the Audience planes button and select plane 1 (the floor) as the audience plane (enter 1 in only the first or both edit boxes and press the Add button). Leave the remaining settings at their default values. Click Save and Run again. The settings-file now instructs the Prediction module to perform ray-tracing with 1 m diameter receivers placed at the Map height (0.5 m) over the audience surface and for each of these receivers to calculate various room acoustic parameters. The result-files created are with sound pressure level in four time-windows (SPL_1K.PLT), Later Energy Fractions in four time-windows (LF_1K.PLT) and D-50, C-80, SPL, G, Ts and LF parameters (PAR1_1K.PLT and PAR2_1K.PLT). For those parameters selected for 3D display separate 3D plot-files (e.g. D-50_1k.PLT will be created). Mapping is only performed for the octave-bands selected (in this case the 1 kHz band) and all defined sources that are selected in General settings are added together taking electrical delays into account if defined (in the source-file).

Early part detailed ISM Switch off Audience area mapping and select Early part detailed ISM and in the Early part detailed ISM dialog select Specular refl. order 3 and I.S. truncation time 100 ms, a Refl. history, and an I.S. space plot-file (I.S. = Image Source). Click Save and Run again. The settings-file now instructs the Prediction module to calculate image sources for the early part of the echogram and create a plot-file for the early part reflections History (H_A0_01_1K.PLT) and an image source Space file (S_A0_01_1K.PLT) for source A0 and receiver 01 @1 kHz. Early part detailed ISM only gives plot-files for the octaves selected and no parameters other than SPL are estimated. This calculation type is to be used for qualitative early specular reflection path analysis or for creating files for the Post-processing and Source addition modules for special cases.

Full detailed calculation Finally, for the full calculation switch off Early part detailed ISM and select Full detailed calculation and in the Full detailed calculation dialog select No of rays/octave 10000, Ray truncation time 1000 ms. Then click Save and Run again. The settings-file now instructs the Prediction module to perform a thorough prediction using randomized tail-corrected cone-tracing (RTC). To preserve high early part detail direct sound, first order diffuse and specular reflections, and second order specular reflections are handled deterministically by the ISM. A plot-file containing a lot of useful information is created (E_A0_01_1K.PLT) (Echogram source A0 receiver 01 @1 kHz). The plot shows full echogram with backward integrated decay, reverberation time regression lines and coefficients, early part discrete reflections, early part smoothed echogram, backward and forward 2 integrated, cos -weighted X (front-back), Y (left-right) and Z (up-down) smoothed echograms. In addition are shown a small hall plan with source- and receiver locations, a dimension scale, source data plus estimated major parameters. Full detailed calculation gives plot-files only for the octaves selected but always calculates for all octaves so that even if an octave-band or specific parameter plot-file is forgotten numbers predicted can always be found in PARAM_ss.TXT (PARAM_A0.TXT in this case) and any selected post-processing files always contain full information for the octave-bands 125 Hz and 16 kHz. This calculation type is to be used for detailed quantitative analysis of an echogram and the results can be used by the Source addition and Post-processing modules (requires selecting the (Save data for) options.

10

1. INTRODUCTION Although, in this example, the three different calculation types were run one by one they can be run in any combination and it is good to always have Geometry view/check selected so that with any changes the latest geometry plot-files always correspond to the latest calculation results. If the hall geometry needs to be changed, the geometry-file must again be edited and the geometry has to be viewed/checked. These were the basic steps running CATT-Acoustic in its most generic form. The Example Hall in Section 2.3 and the Tutorial in Section 11 give more realistic examples. Note that the Tutorial can first be “run” through directly in the help-file since many created plots are shown.

1.3 Common user interface elements This section describes user interface elements (menus, dialogs, toolbars etc.) that are common to all seven modules in CATT-Acoustic.

Menus When a menu item is high-lighted a description is displayed in the left status-bar field of the main window. Common menu items are found in the File, the Window, and the Help menus. File|Open Recent file-type | file-names All modules, except Surface properties, have a menu item where the ten most recent settingsfiles etc. are stored for direct selection. External CATT Tool| CATT tools that are external to CATT-Acoustic (OpenGL 3D-Viewer, PLT-Viewer, DXF2GEO, CATT-Walker etc.) can be loaded from this menu. At program startup the CATT folder is searched for available tools and those found are added (original file names only, any renamed tools will not be added). Also the free GratisVolver and GratisWavPlayer will be listed here if found. File|Preferences… Displays the Preferences settings dialog. Before displaying the dialog the current CATTDATA-folder can be confirmed or changed. The Preferences settings applies to the current CATTDATA-folder. Note: Hidden options, documented further below in this section, act as another set of preferences for less common options.

11

1. INTRODUCTION

Projects folder

Preferences

The parent folder under which you wish to place your projects. This folder is then the startup folder for folder-selection with e.g. Utilities|Create a New Project and initial browsing for input folder in General settings (when the Input folder field is empty).

12

1. INTRODUCTION Surface properties

Preferences

Library used in the Surface properties module and when geometry is read. Libraries are placed in the CATT-folder ABSLIBS. Default anechoic WAV-file

Preferences

Used for convolution. If a favorite file is used it is convenient to enter it here since it will be preselected in the Convolution dialog. Anechoic WAV-files are placed in the CATTDATA-folder ANECHOIC. HRTF library

Preferences

HRTF library used for binaural post-processing at each samplerate. Demo auralization HRTFs for 44.1 kHz, based on a rigid sphere approximation, are named SPHERE_plain_44.DAT. For HRTFs other than those originally provided check the CATT users’ www page. HRTF library-files are placed in the CATTDATA-folder HRTFS. Headphone filter

Preferences

Headphone filter used for binaural post-processing at each samplerate. Demo auralization filters for 44.1 kHz are named GENERIC_V8_plain_44.DAT. For filters other than those originally provided check the CATT users’ www page. Headphone filters are placed in the CATTDATA-folder HEADPHONES. CATT-Edit font

Preferences

The font used by CATT-Edit when it is loaded from the main application. Don't pay any attention to the actual font-size number selected, just select a font size that works well. When CATT-Edit is loaded from e.g. the Windows Explorer, a default font is used. Warn if settings may be inappropriate

Preferences

If potentially inappropriate settings have been selected, warnings will be issued and an option to abort the action is presented. With sequence (batch) runs warnings are temporarily switched off. Experienced users may want to set some warnings off. No direct sound (Prediction)

Warning options

Warning if for some source-receiver combinations the direct sound is blocked. Note that most measures loose their meaning if the direct sound is not present and predictions will be less accurate. Too few rays used (Prediction)

Warning options

Warning if the selected number of rays is less than half the number of rays estimated by "auto number". Too short echogram (Prediction)

Warning options

Warning if the ray truncation time is selected too short. If so T-30 will not be correctly estimated.

13

1. INTRODUCTION Head direction source (> 1 source) (Prediction)

Warning options

Warning if more than one source is used and the head direction in Prediction:General settings is set as "source". This is not a meaningful setting except when several sources are calculated as independent cases and the sum of the results at each receiver is never considered. Overwriting old output results (Prediction)

Warning options

Warning if the output folder already contains such files that the current calculation will create. Diffuse reflections off (Prediction)

Warning options

Warning if diffuse reflection is switched off in Prediction:General settings. Diffuse reflection should normally always be on and surface properties should include scattering coefficients. Default diffusion off (Prediction)

Warning options

Warning if the default diffusion is switched off in Prediction:General settings. Default scattering coefficients are assigned to all surfaces that are not given specific coefficients. ISM diffuse reflection order = 0 (Prediction)

Warning options

Warning if diffusion is on but in Early part detailed ISM the diffuse reflection order is set to 0. Too short impulse response (Post-proc. related)

Warning options

Warning if the impulse response in Post-processing:General settings is too short in relation to the estimated RT. Some early reflections excluded (Post-proc. related)

Warning options

Warning if the early part reflection min order is > 0 (full Huron version only). Post-processing based on ISM (Post-proc. related)

Warning options

Warning if post-processing files are created using Early part detailed ISM, normally they should only be created by Full detailed calculation. Long calc. no post-processing files (Post-proc. related)

Warning options

Warning if a long prediction is started and some of the post-processing options are not selected. Use a small Plot-file control

Preferences

Uses a small plot-file control (instead of a bigger one) for navigating 3D views etc. in the Viewmodule. Most controlling functions can be reached via the toolbar or via direct mouse operations. Show Plot-file control

Preferences

Opens the plot-file control when the first plot-file after startup is viewed. Most controlling functions can be reached via the toolbar or by direct mouse click-drag on the window so the control is rarely needed.

14

1. INTRODUCTION Auto minimize View-module

Preferences

Causes the View-module to minimize as soon as another module is activated. Useful if run in low screen resolutions such as 640x480 VGA (at least 800x600 is recommended). Disable screensaver while calculating

Preferences

Some screensavers take a lot of CPU time and can slow down the calculations by almost an order of magnitude and therefore often have to be disabled. This option disables any screensaver only during calculation and is best kept selected on at all times. 2DA ISM History plot (instead of 3DA)

Preferences

The Early part detailed ISM reflection history plot (H_ss_rr_oct.PLT) is normally made as a 3DA plot but the older 2DA version can be switched on (e.g. so that big complex rooms can be zoomed). Thus, either 2DA or 3DA can be selected but not both. Calculate C-50/dB (instead of D-50/%)

Preferences

Some users prefer C-50 in dB rather than D-50 in % (C-50 is calculated the same way as C80 but with a 50 ms limit instead of 80 ms). This is how the two measures (that express the same thing) map to each other for every 10% in D-50 (values < 0.1 and > 99.9 will in the table below be mapped to -30 and 30 dB respectively). D-50

0

10

20

30

40

50

60

70

80

90

100

%

C-50

-30

-9.5

-6.0

-3.7

-1.8

0

1.8

3.7

6.0

9.5

30

dB

Thus, either C-50 or D-50 can be used but not both. Depending on the type of calculations there are some differences: •

Full detailed calculation: The type of values displayed cannot be changed afterwards by changing the Preferences option. Note: the RASTI_ss.PLT file always shows D-50 since the main reason is to show that the much (much) simpler measure D-50 gives just about the same numbers as the more complex RASTI does.

•

Audience area mapping: The type of values displayed can be changed afterwards by changing the Preferences option and adjusting the Options|Map Scale in the Plotfile viewer. However, the stand-alone PLT-viewer always shows D-50 in maps (this may change later by adding a special D-50/C-50 option to the viewer).

•

Source addition: The type of values displayed cannot be changed afterwards by changing the Preferences option.

Create 32-bit SIM-files (not for Lake h/w)

Preferences

With this option off (default) created SIM-files are scaled optimally for use with Lake's 24-bit convolvers. However, with the CATT-Acoustic convolver and The FIReverb Suite MultiVolver full 32-bit coefficients may be used. This is mostly interesting if filters are created in CATTAcoustic (e.g. for a 5-channel to binaural down-mix) but will be used in the MultiVolver that also can use and create 24- and 32-bit WAV-files since for 16-bit WAVs the current SIM-file scaling is well sufficient. Use CATT syntax for HeadScape SIM-files

Preferences

Only shown for full version with Lake Huron add-ons. With v7, the only application of HeadScape files was with Lake DSP hardware. With v8 they can also be used with the Walkthrough Convolver. With this option selected, the files name syntax of HeadScape files

15

1. INTRODUCTION will be E_ss_rr_hhhvvv_c.SIM rather than Hhhhvvvc.SIM where c is "L" or "R", hhh the horizontal angle and vvv the vertical angle. Graph color

Preferences

Sets the background graph color. The selections are quite few (only light colors will work). One use of this could be to change color for a new hall version (as well as a new output folder) to make comparisons clearer. Note that the selection affects only graphs created after a change. Default time-constants

Preferences

Default time-constants for echogram smoothing. The values entered here will be set for any Prediction:File|New Settings and also for the initial setting when the Prediction module is opened. The default values apply both to Full detailed calculation echograms and Early part detailed ISM Sone-scale echograms. Note: for "sum" echograms the 1k Hz value is used. STI weights

Preferences

Selection of bands and weights for STI (STI = Speech Transmission Index): Original

Normal “classical” STI weights [Steeneken-67]

STI modified

Weights and bands according to Floyd & Steinberg and as measured by MLSSA [MLSSA] (in addition to original STI and RASTI), see Prediction Theory 2.5. These values are based on the Articulation Index (AI) weights from F&S (AI is a predecessor to STI but took into account only the S/N ratio in each band and no modulation, see e.g. [Steeneken-67]). It appears that MLSSA is the only measurement system where these weights are used as an option.

User

user settings: sum of weights for all used bands most be 1.0.

IECmale/female

STI for male or female voice according to CEI/IEC 60286-16, 2003-05 [IEC 2068-16]. Includes the redundancy correction β factors, level dependent masking and absolute threshold. For experimental purposes, the latter two can be disabled via a not official hidden option, contact CATT for the code.

16

1. INTRODUCTION Include in overview text-files (8 and 16 kHz options)

Preferences

If 8k and 16k Hz results should be given in output text-files such as PARAM_ss.TXT. This setting also controls if 8k and 16k Hz are shown in the Prediction:Interactive RT estimate dialog. Include in overview plot-files (8 and 16 kHz options)

Preferences

If 8k and 16k Hz results should be given in output plot-files such as E_ss_RR_INT.PLT and RT_A0.PLT. Include in map 'sum' (8 and 16 kHz options)

Preferences

If audience area mapping 'sum' is selected, all bands 125 - 16k Hz will be calculated. With this option off, only 125 - 4k Hz will be calculated and the sum is not affected by results at 8k and 16k Hz. Which bands are used is indicated on map sum plot-files. Initial view angles (3D-balloon)

Preferences

The initial view angles for directivity balloon plots in the Directivity module can be selected. The horizontal angle goes from 0° to 360° where 0° is at the front and 90° is from the loudspeaker's left. The vertical angle goes from -90° to 90° where 0° is in the horizontal plane and 90° is from above. SD2 2° (3D-balloon)

Preferences

Will plot SD2 balloons with a 2° resolution, else 5°. Shade (3D-balloon)

Preferences

Will color-shade 3D-balloons. Lines (3D-balloon)

Preferences

Will include grid lines on color-shaded 3D-balloons. SPL range for map statistics

Preferences

The default SPL range for MAPSTAT.TXT (and MAPSTATx.TXT) is 20 dB (corresponding to max..max-20 dB), the range is here selectable between 5..60 dB. File|New

This menu has a toolbar equivalent where applicable

Creates a new settings file (e.g. a new Prediction settings-file in the Prediction module). File|Open...


Opens an existing settings file (e.g. a Prediction settings-file in the Prediction module). File|Save...


Save the current settings file (e.g. a Prediction settings-file in the Prediction module). File|Save As... Save the current settings file (e.g. a Prediction settings-file) under a new name.

17

1. INTRODUCTION File|Edit Text-file... Loads a selected text-file into CATT-Edit for general purpose editing or print-out. The Prediction module has (Edit) buttons on its main dialog and in General settings for faster access to project files. Note: the External CATT Tool menu can start CATT-Edit with a new empty file. File|View Text-file... Loads a text-file into CATT-Edit in read-only (view) mode. File|Close Module Closes the active module. Prompts for saving data if required. File|Exit Terminates CATT-Acoustic. Prompts for saving data if required. Window|Available Modules list Lists all modules and check-marks those currently open. By selecting an unchecked item the module is opened and by selecting a checked item, the corresponding module is closed. Prompts for saving data if required. Window|Cascade Lines up the modules from the upper left corner of the main window. Window|Arrange Icons Arranges minimized module icons (may be required if the main window size has been changed). Window|Close All Closes all open modules but does not exit the program. Prompts for saving data if required. Window|Open Modules list Lists currently open modules. By selecting an item the corresponding module is activated. This menu has a toolbar equivalent since open modules are shown as toolbar buttons. Help|Help on Current Module Displays the first help-file page on the currently active module. F1 gives context sensitive help. Help|Contents Displays the help-file Contents topic. Help|Search for Help on... Displays the help-file keyword search engine. Help|View Manual Addendum

18

1. INTRODUCTION A direct link to an addendum file in the CATT-folder. This file in included in all updates and documents additions and changes. The help-file is always up to date, only the manual needs the addendum. Help|About CATT-Acoustic... Presents an "about box" with copyright, license name, license expiry date, current operating system notices and a direct link to the users’ web page.

Dialogs In many dialogs, small square push-buttons without labels selected. These are used for editing or specifying the item further.

appear when an item is

Dialog Help button or F1 give context sensitive help (also if a dialog has no Help button).

Toolbars Many basic menu functions can be controlled via the toolbar:

New Open Save

Prediction Post-processing Source addition Surface properties Directivity Sequence processing Plot-file viewer

At the left side of the toolbar the standard "New", "Open", "Save", and "Print" buttons (when applicable) are shown. In addition, currently open modules have a button each making it easy to switch between open modules. To open or close a module the Window menu is used. The same image is used for each module's window title icon as well as for file-types belonging to the module. The Plot-file viewer module has an extensive toolbar duplicating most controls on the optional floating Plot-file control (Ctrl+R) dialog:

The toolbar tooltips displays the current value for a 3D setting (the current camera position and target for camera mode, i.e. the aim line end-points as seen in the Camera Set mode). The Post-processing module has further toolbar icons for the most commonly used Utilities menu items:

19

1. INTRODUCTION

Hidden options To be able to meet special requests without generating many special software versions, (hard to keep track of, support and update) a feature called "hidden options" was introduced with v8. Many times a special request may be easy to meet and may take up little space in the EXE but may be of interest to only one or a few users. Rather than to clutter the interface with options very seldom used, these are added as hidden options that can be unlocked or switched on/off with unique code keys. It may also happen that a hidden option will be converted to an open option and be included in the normal interface and it can therefore also serve as a way to test new things in an orderly fashion. Similarly, a little used option may be moved to be hidden for those that still want to use it. Note that these are typically not "secret" options but just very unusual options like e.g. special export files needed for a particular project only. There is no practical limit to the number of options that can be handled this way. The procedure is simple: When CATT32.EXE is started it looks in the CATTDATA-folder for a file called HIDDENOPTIONS.TXT, if it exists it reads codes from it (one per line) and if these codes are valid codes the corresponding hidden options are enabled. Codes can be disabled by commenting them out (with a semicolon) but for the change to take effect the program has to be restarted. The first two open codes correspond to removed v7 features: Include a map of all planes: 45BCE-72843-0537E-98977 Causes a map of all planes used to be included in output text files (E_ss_rr.TXT, PARAM_ss.TXT). Planes with some diffusing properties are marked by an asterisk. Create an ISM text history file: F210A-C20D6-A4EBB-C0493 Causes a text history file (I_ss_rr.TXT) to be created with Early part detailed ISM. This file is documented in the Prediction output files section (2-4). Show all defined corners: 71E2E-5CBA4-1F06C-133DA Normally in COINFO.PLT only corners actually used in a plane are shown, this code will instead allow all defined corners to be shown. Corners not used in any plane definition are shown in red. Switch off contours: AC5D6-3EC76-0A444-9B627 When using SD2 array sources that do not have the main lobe on the source axis, contours are not so meaningful and take time to calculate. This code switches off all contours no matter if functions with contours are selected or not. STI/RASTI even with direct sound only: 249BB-4604A-DF146-07BD8 For an outdoor arena such as a football stadium the STI/RASTI of the PA systems is often determined by late arrivals from neighboring speakers and background noise and not to such a high degree by reflected sound. For cases where the user judges this to be the case, a hidden option enables mapping of STI and/or RASTI using only the direct sound. This option must be used with caution but can potentially speed up initial system design considerably. Do not plot high plane edges in maps: 07D55-F67D8-3515B-957E9 Normally in audience area mapping, all edge lines are drawn but if a very detailed ceiling is used the maps can be too cluttered with lines and hard to read. This hidden option will cause lines that have both ends above the map not to be drawn.

20

1. INTRODUCTION Option to plot only the audience planes in audience area mapping: DA756-28EF0FA33B-083D5 In rooms where the actual audience area surface is much smaller than the room, this hidden option can be used to plot only the audience surfaces thus giving a higher graphical resolution. Option for finer RASTI/STI map statistics ranges: 4825C-27E27-A141D-284B6 The MAPSTAT.TXT file normally shows RASTI and STI values in ranges corresponding to "GOOD", "FAIR" etc. This option will instead use finer 5% ranges. Option for mapping of closest source group: CF553-70024-CF5D7-9D8A3 The "Closest" audience area mapping plot (direct sound mapping only), previously (< v8.0e) indicated the letter of the source group with the shortest delay instead of a color as indicated in a color scale. The old group letter map is available via this hidden option. Option to show even very low STI values: FEA33-0CCB5-A8D9A-1663D From v8.0c Build 6, very low STI with noise values (BAD,POOR) were not shown since it sometimes gave problems. This hidden option allows all STI to be shown. Options to calculate U50, AI, PI, SII A number of less usual measures are calculated by Full detailed calculation only and given in result text-files when selected. Before using any of these options it is strongly advised to read the papers and standards cited to judge their applicability. In most cases STI will work as well or better. U50: 1938B-AFFA4-2C280-F2F0E I.e. C50 with background noise taken into account, see [Bradley-96:2]. AI, Articulation Index: B506E-AE71E-A447B-DECD0 See e.g. ASTM E 1130. PI, Speech Privacy Index: B58F0-2CA5C-FA209-808C8 Calculated as 100 (1 - AI) [%], see e.g. ASTM E 1130 for an application in open offices. SII, Speech Intelligibility Index: B33E1-779A1-BE75E-A0F79 See ANSI S3.5-1969. This is a very complex measure and only the version that does not take into account room reverberation has been implemented (standard section 5.1). Option to show helper curves on echograms: C56DF-2DB1E-234B6-74F60 On the early Full detailed echogram in E_ss_rr_oct.PLT and in the Early part detailed ISM history plot H_ss_rr_oct.PLT three curves can optionally be displayed to help evaluate the importance and audibility of reflections in some contexts (such as small rooms). The curves used come from Fig 24 in [Toole-90]. The upper curve is related to the audibility of a "Separate Echo" (Lochner & Burger) . The middle curve is related to "Image Shift or Spreading" (Olive & Toole) The lower curve is related to "Absolute Threshold" (Olive & Toole) Before using this option it is strongly advised to read the paper to judge the applicability.

21

1. INTRODUCTION Create MATLAB output from Full detailed calculation: A677D-603DB-876F8-56C5D This option will create a [MATLAB] v4 format version of the PARAM_ss.TXT result file called PARAM_ss.MAT. The variables can optionally be prefixed with the string “catt_” as selected via another hidden option, see below. The reason for this option is that with later MATLAB versions, all variables can be read into a struct e.g. like this: catt = load(['c:\temp\out\param_a0.mat']); where a variable can then be referred to as catt.variable with no need for a prefix, the variables (the struct field-names) can if needed be read via fieldnames(catt). One of several uses for a MATLAB output is with the Exec command interface of the Sequence processing Prediction step that can run a MATLAB script afterward completed prediction. Below are listed the saved variables in alphabetical order as when using the whos command with variables starting with a capital listed first and shown without the optional prefix. Most names are obvious and the same as in PARAM_ss.TXT. nrec is not an actual created variable but must be calculated using nrec = length(rec_ids). Arrays with x 8 format give values for 125..16kHz, x 9 also gives the sum for all octaves, x ? is case dependent. AI AbsC AbsCg Background C50 C80 D50 EDT EyrT EyrTg G LF LFC PI RASTI ResidBackground S SII SPL STI STI_name SabT Sact T15 T30 TI_nn TI_wn Ts U50 V Vact air_density air_dissipation air_humidity air_temperature build diffs mfp orig_filename orig_savetime project rec_ids

1 x 1 1 x 8 1 x 8 1 x 8 nrec x nrec x nrec x nrec x 1 x 8 1 x 8 nrec x nrec x nrec x nrec x nrec x 1 x 8 1 x 1 nrec x nrec x nrec x 1 x 10 nrec x 1 x 1 nrec x nrec x nrec x nrec x nrec x nrec x 1 x 1 1 x 1 1 x 1 1 x 8 1 x 1 1 x 1 1 x 1 1 x 8 1 x 8 1 x ? 1 x 19 1 x ? nrec x

9 9 9 8

9 9 9 8 2

1 9 2 8 8 8 7 7 9 9

1

Articulation Index *) % % Background SPL, dB dB (if not D50 is chosen in Preferences) dB % (if not C50 is chosen in Preferences) s s s dB % % Privacy Index *) (:,1) is without noise, (:,2) with noise, % Residual background SPL, dB 2 Surface area, estimated by rays, m Speech Intelligibility Index *) dB (:,1) is without noise, (:,2) with noise, % STI type name **) s 2 Surface area, geometrically calculated, m s s Transmission Indices without noise, % Transmission Indices with noise, % ms dB *) 3 Volume, geometrically calculated, m 3 Volume, calculated by rays, m 3 kg/m -1 m % °C CATT-Acoustic build number: build*100 + subbuild Mean scattering coefficient, % Mean free path, m Original file-name **) Original file save date and time **) Project name **) Receiver ids

22

1. INTRODUCTION rec_ids_ST rec_loc soundvelocity src_aim src_delay src_directivity src_dirindex src_id src_loc src_lp1m src_roll version

? x 1 nrec x 3 1 x 1 1 x 3 1 x 1 1 x ? 1 x 8 1 x 2 1 x 3 1 x 8 1 x 1 1 x 19

Ids of the receivers with Stage Support values Receiver coordinates, m m/s Source aim coordinates, m Source delay, s Source directivity name **) Source directivity Index, dB Source id **) Source coordinates, m Source level at 1m on axis, dB Source roll angle, º CATT-Acoustic version **)

*) if the corresponding hidden option is selected **) strings are stored as character ASCII values, use char(variable) to convert to a string variable Note: if a parameter value could not be calculated (say C80 if no sound exists after 80 ms) the value will be -1000.0 (no parameter will take that value). Prefix variables in MATLAB output: AC8D1-BD42D-4B048-7856C See discussion above about prefixing variables with “catt_” (option on) or not. Automatically run a sequence : 9B237-61F2C-499B9-82935 With this option a Sequence-file (.SEQ), given on the command-line to catt32.exe (enclosed in " " if the path contains spaces) or double-clicked in Explorer, will automatically run the sequence and after completion close down CATT-Acoustic. The purpose of this option is to be able to run several independent sequences in batch: c:\catt\catt32 c:\hall1\full.seq c:\catt\catt32 c:\hall2\full.seq c:\catt\catt32 "c:\new hall\full.seq"

Miscellaneous If the Plot-file Viewer module is open, plot-files created are shown automatically after e.g. a prediction. If it is closed, the same results are created but the Viewer module first has to be opened to be able to see created plot-files. After the module is opened select List|Open Listfile and load either PLT_GEO.TXT, PLT_GEO_FULL.TXT, etc. depending on Prediction combination, or MULTPLT.TXT or POSTPLT.TXT depending on viewing result from the Source addition or Post-processing module. When the Sequence module is used plot-files are not shown automatically, even if the View module is open but can be viewed afterwards, see Section 8. The progression bar shown in the Processing dialog, while performing e.g. a prediction, displays the estimated remaining processing time for each processing step. Note that the image source calculation in Early part detailed ISM is a non-linear process and the estimated remaining time estimate may fluctuate considerably. The main application window can be minimized, moved or resized during calculation to allow e.g. for background calculation as an icon. When result text-files are created their names show up in a small list-box and can be selected for viewing. The files are then loaded into CATT-Edit in view-only mode, with some menus

23

1. INTRODUCTION disabled, so that results are not accidentally altered. Text can still be printed out and copied to the clipboard. In view-only mode the editor window title displays "CATT-Edit (view)". The current settings-file name for a module is shown with full path in the left statusbar when the module is active. CATT-Acoustic uses the Windows ANSI character set exclusively so if e.g. a GEO-file has been edited in a DOS editor care must be taken when using upper-ASCII characters. CATTEdit has an option to convert from the MS-DOS OEM to the Windows ANSI set if required. The interface uses the decimal character selected in the Control Panel:Regional settings so that if e.g. decimal comma is selected it must be used in all edit boxes in the interface and it is used for all displayed decimal values and for all created results (with the exception of CADfiles for the AutoCAD export and WRL-files for VRML export that always use decimal point). Input text-files may use either decimal point or decimal comma independent of the settings in Control Panel (except CATT-SD0 and CATT-SD1 directivity import-files and HTR-files for head-tracking that may be comma-delimited and thus only can use decimal point). All files used by CATT-Acoustic can have long filenames except source directivity files (*.SD0, *.SD1, *.SD2, *.CF1, *.CF2) that can be max 15 characters and no spaces (including a sub-folder name and “\”) plus extension and HRTF libraries and headphone eq filters (*.DAT) that can be max 40 characters long plus extension. Dialog Help buttons and F1 give context sensitive help (also if a dialog has no Help button). CATT32.EXE is registered as the "Open" application in Explorer for CATT-registered files when double-clicked (Explorer:View|Options... menu). File extensions handled by CATT32.EXE are: PRD PST PLT SEQ SD0, SD1, SD2 GEO, LOC, CAD, HTR

loads the PRD-file for the Prediction module loads the PST-file for the Post-processing module loads the PLT-file for viewing in the Viewer module loads the SEQ-file for the Sequence processing module Note: a hidden option allows for an automatic run of a sequence loads the SDx file for the Directivity module loads CATT-Edit. Note: Do not assign CATT-Edit for this task.

UNCTT.EXE is registered to automatically load CTT-files. CATTOGL32.EXE is registered to automatically load OGL-files. Remaining CATT-registered file-types have to be loaded within a CATT program and are ignored if double-clicked. If a processing is currently under way a double-click is also ignored. Because of the way Win32 handles file types (relying only on the extension) it may happen that an extension used by CATT clashes with an extension used by another software. Because of the primitive mechanism used by Win32 it is the latest installed program that controls the file type. If you have any problems with any of the file-types being assigned to CATT-Acoustic, contact CATT for details on how to change the reference. Temporary files are created in the folder for temporary files as (if) given by Windows. Big temp-files are used mostly by the convolver and for WAV- and SIM-file additions, scaling and calibration. The CATTDATA-folder ANECHOIC stores temporary PPW-files used by the convolver for speed reasons, these files can be deleted at any time after a completed convolution.

24

1. INTRODUCTION

1.4 Input/output files overview *.DAT Surface properties (ABSLIBS)

*.SD0,*.SD1,*.SD2,*.CF1,*.CF2 (

Geometry, Sources, Receivers:

Prediction

E_ss_rr.ADD SRC.CMB

MASTER.GEO REC.LOC SRC.LOC

Source directivity

)

Multiple source addition source combination, directivity, aim, eq/gain, delay

*.PLT *.TXT *.OGL

*.PLT *.TXT

E_ss_rr.ECH

E_ss_rr.ECH Walker.CAG Post-processing module HRTFs (HRTFS)

*.DAT

Headphone eq (HEADPHONES)

*.DAT

Post-processing source eq/gain (on/off)

*.PLT *.MAT Export files *.MLS

E_ss_rr_f.SIM SIM-file addition source combination, gain, delay ????_rr_f.SIM

Anechoic material (ANECHOIC)

*.WAV

Softw are convolution

*.WAV *.SCL

WAV-file addition source combination, gain, delay *.WAV

Fig. 1.2 General data flow. E_ss_rr_f.SIM is shorthand for the filename syntax used (e.g. in E_A0_01_L.SIM and E_A0_01_R.SIM for binaural responses) where ss is a two-character source id and rr is a two-character receiver id. CATTDATA-folders are shown in round brackets. Description of input and output files created and required by the various modules in CATTAcoustic, for reference purposes. W and R are used to mark if the module writes (W), reads (R) or both writes and reads (W/R) the file-type. Small ss and rr denote source (A0..Z9) and receiver (00..99) ids and oct the octave-band (125, 250, 500, 1K, 2K, 4K, 8K, 16K, sum). All files except *.TXT, *.TX1, *.TX2, *.SD2, *.GEO, *.LOC, *.WCS, *.HTR, *.WRL, *.CF1, *.CF2 and *.CAD are in internal CATT binary format. Fig. 1.2 illustrates the general data flow between modules. For CATT-Walker also a file Walker.CAG is created along with E_ss_rr.ECH and postprocessing for CATT-Walker created files called name_ss.CWI where name is the name of the selected CAG-file, for details see Walker.PDF.

25

1. INTRODUCTION

Source directivity module *.SD0,*.SD1,*.SD2 source directivity library-file(s) in the CATTDATA-folder SD or sub-folders (W/R) *.CF1,*.CF2 CLF source directivity library-file(s) in the CATTDATA-folder SD or sub-folders (R) [CLF] *.COL

directivity collection files (W/R)

*.TXT, *.TX1, *.TX2 CATT-Acoustic text file-format for 15° and 10° directivity formats (CATT-SD0 and CATT-SD1 respectively) (W/R) *.PLT

various result plot-files (W)

*.MRK

marker files for source 3D visuals (R)

Surface properties module *.DAT

surface (absorption/diffusion) properties library-file in CATTDATA-folder ABSLIBS as selected in Preferences (W/R)

*.DAT

Creation of a new generic properties library in CATTDATA-folder ABSLIBS (W)

*.TXT

CATT-Acoustic extended text library exchange file-format (W/R)

Prediction module *.PRD

settings-files (W/R)

Geometry-related files in a folder chosen to hold the hall data, the "input" folder: *.GEO

geometry-file(s) describing the hall to model (R)

*.LOC

receiver location file, source location/data file (R)

*.MRK

marker files for non-acoustical visual elements (W/R)

Library data in the CATTDATA-folder SD or sub-folders : *.SD0,*.SD1,*.SD2,*.CF1,*.CF2 source directivity files (R) Library data in the CATTDATA-folder ABSLIBS: *.DAT

absorption/diffusion prop. library as selected in Preferences (R)

Various result-files: *.PLT


26

1. INTRODUCTION SHADED.OGL

a shaded 3D model for the CATT 3D-viewer (W)

PLT….TXT

list of the latest result plot-files depending on Prediction functions selected (W)

E_ss_rr.TXT

single receiver result text-file from Full detailed calculation (W)

PARAM_ss.TXT

multiple receiver result text-file from Full detailed calculation (W)

PARAM_ssX.TXT

multiple receiver result text-file for MS Excel import (W)

*.CMB

source combination data for source addition (W)

E_ss_rr.ADD

compact echogram results created for source addition (W)

E_ss_rr.ECH

compact echogram results created for post-processing (W)

Walker.CAG

room geometry for CATT-Walker.

*.CAD

intermediate export file format for the AutoCAD interface (W)

*.WRL

VRML 2.0 file export of the hall geometry (W)

*.DXF

DXF file export of the hall geometry (W)

DEBUG.TXT

Data for geometry debug purposes (W);

Master_tree.TXT GEO-file tree for debug (name derived from current Master-file) (W); Master_RT-estimate.TXT Interactive RT estimate result-file (W); Master_stat.TXT Interactive RT estimate result-file (W);

Source addition module This module writes and reads all data in the folder where the CMB-file is read. *.CMB

source combination data settings-file (W/R)

E_ss_rr.ADD

echogram results created by the Prediction module (R)

E_ss_rr.ECH

echogram results created for post-processing (W)

E_rr.TXT

single receiver multiple source result text-file (W)

E_nn_rr_oct.PLT

single receiver multiple source echograms (nn = 01,02,...) (W)

TREND 1, 2, 3 .PLT

multiple receiver multiple source result plot-file (W)

PARAM.TXT

multiple receiver multiple source result text-file (W)

MULTPLT.TXT

list of the latest result plot-files (W)

27

1. INTRODUCTION

Library data in the CATTDATA-folder SD or sub-folders: *.SD0, *.SD1, *.SD2,*.CF1,*.CF2 source directivity files (R)

Post-processing module This module writes all results to the folder where the input ECH-file is read. The module can also perform file conversion between several formats such as SIM, TIM, MAT, WAV, and 16bit integer sound-files. *.PST

post-processing settings-file(s)

*.ECH

compact echogram results created by the Prediction and the Source addition module (R)

*.CAG

room geometry and other info required for post-processing for CATT-Walker (R)

*.SIM

created impulse response data in Lake SIM-format [Lake] (W)

*.CWI

created impulse response data for CATT-Walker (W)

*.MLS

created impulse response data in MLSSA ASCII format (W)

*.MAT

created impulse response data in MATLAB v4 [MATLAB] format (W)

*.HTR

head-tracking angle description file for the Lake DSP HeadScape application (only Huron/CP4 add-on version [Lake]) or with the Walkthrough convolver (R)

*.PLT


POSTPLT.TXT list of the latest result plot-files (W) Software convolution: *.SIM

created impulse response data in Lake SIM-format (R)

*.WAV

anechoic data (R and plays)

*.WAV

processed (convolved) data (W/R and plays)

*.SCL

scaling data for relative calibration (R/W)

*.WCS

Walkthrough Convolver Scripts (R/W)

Library data in the CATTDATA-folder ANECHOIC: *.WAV

anechoic files for convolution (R)

*.AWI

anechoic WAV-file info (R/W)

Library data in the CATTDATA-folder HEADPHONES: *.DAT

headphone equalization filter as selected in Preferences

28

1. INTRODUCTION Library data in the CATTDATA-folder HRTFS: *.DAT

HRTF library as selected in Preferences

Plot-file viewer module *.PLT

plot-files (W/R)

*.EMF

Windows Extended 32-bit Meta-files export (W)

*.BMP

Windows Bitmap-files export (W)

*.TXT

text-file lists of plot-files (and associated WAV-files) (R/W)

*.WAV

WAV-format sound-files (R and plays)

Sequence processing module This module, in addition to its own settings-files, reads/writes a variety of files according to which module is being executed in the script. Even if SEQ-files mainly contain text information, they are in binary format and must not be edited. *.SEQ

sequence script-files (R/W)

1.5 Software overview CATT-Acoustic is a room acoustic prediction program based on the Image Source Model (ISM) for early part echogram qualitative detail, Ray-tracing for audience area color mapping and Randomized Tail-corrected Cone-tracing (RTC) for full detailed calculation enabling auralization. The system consists of a 32-bit Windows MDI (Multiple Document Interface) main program, a customized Notepad-like text editor that communicates with the main program, a stand-alone 3D-viewer based on OpenGL, a stand-alone PLT-viewer, the external DXF2GEO and other utilities, CATT-Walker, WalkDecim and a set of AutoLISP files for the optionally used AutoCAD interface. Additional external modules for real time walkthrough convolution (CATTWalker), DXF conversion (DXF2GEO) and some other tasks can be loaded from File|External CATT Tool. The main program CATT32.EXE integrates prediction, postprocessing for auralization, software convolution, addition of multiple sources, source directivity management, surface properties library management, sequence (batch) processing, WAV-file playing, file conversions, plot-file viewing/printing and several other functions.

Hardware requirements CATT-Acoustic requires an IBM PC compatible equipped with a Pentium processor and Windows 95, 98, ME, NT 4.0, 2000, XP or Vista. For auralization a soundcard capable of 16bit 44.1kHz stereo replay is required.

29

1. INTRODUCTION

Hall geometry and absorption The hall geometry can have any shape as long as it can be approximated by a maximum of 99,999 planes (plane surfaces) although severely convoluted shapes (such as L–shapes or worse) will give less good prediction results. This limit is quite arbitrary and can be increased in the future, if required, but generally better results are not created by a very detailed model. As the geometry input is made by using a text editor (although other options such as via DXF2GEO or via AutoCAD or other modeling plugins exist), the input-file format has been made very forgiving allowing for blank lines and comments and no need to place the data in fixed columns. Powerful facilities such as symbolic constants, expressions, calls to math functions, IF-statements, tracing statements, interactive input with data checking, and hierarchic files for the geometry are incorporated. Rather then just read by the main program the geometry-files are interpreted. Overview of the geometry description: ·

number of corners/nodes in the model (in practice) only limited by available memory

·

hierarchic geometry-file organization enabling structuring of the geometry: main hall balconies reflectors extra reflectors audience

·

geometry-files can be scaled e.g. to compensate for erroneous drawings or to convert from imperial to metric units, from mm to m etc.

·

a geometry-file can: ·

be defined as an object and then be translated and/or rotated (typically for reflectors)

·

be defined as an object and be made to duplicate its contents with a new set of translation/rotation values

·

mirroring of symmetrical (or nearly symmetrical) parts of a hall

·

symbolic constants and expressions with calls to math functions can be used to specify coordinates, translations, rotations and several other values

·

symbolic string constants

·

functions for creating model corners that are automatically locked to planes, and for creating corners as intersections between lines and planes

·

tools for maintaining a structured model, such as declaring e.g. the x-value of a corner to be the same as the x-value of another corner, or the use of local corner and plane numbers in each file.

·

planes can be sub-divided in any number of sub-planes each with different absorbing/ diffusing properties. By dividing a plane consisting of several different absorbing/diffusing parts into sub-divisions instead of making one plane definition for each part, the model will be more logical and the calculation time will decrease.

·

planes and plane sub-divisions can be concave (i.e. have re-entrant angles) or convex with number of corners and number of sub-divisions (in practice) only limited by available memory

·

plane corners can either be entered ordered clockwise as seen from the back side or from the front side of reflecting planes (user choice but must be consistent in each GEO-file).

·

libraries of named absorbing/diffusing surface properties. Number of entries only limited by available memory.

·

absorbing/diffusing properties can be specified in several ways:

30

1. INTRODUCTION

·

·

as a named library entry (very common materials as entered in the Surface properties module)

·

as a named property in a geometry-file (common materials in a certain hall)

·

by complementing an absorption-only library material with scattering coefficients when used in a GEO-file

·

directly in % in a plane definition (not so common materials that perhaps need no names)

·

a plane or a plane sub-division can be assigned an automatic frequency dependent edge diffusion to emulate the diffusing effects of diffraction

·

frequency dependent default scattering coefficients can be specified.

marker elements or loaded marker-files for non-acoustical visual elements.

Coordinate system In principle any right-handed coordinate system could have been used, but to use mirroring of symmetrical parts and for view angles for perspective plots, aiming angles for a source, reflection incidence angles and head-direction Stage to be understood, the system has to be as follows, see Fig. 1.3. Imagine standing on the stage of the hall looking towards the audience: ·

The x-axis should run from left to right

·

The y-axis should run towards the audience

·

The z-axis should run upwards.

z

stage side view

y

y x

top view

Fig. 1.3 Hall coordinate system

The origin can be placed anywhere but it is recommended to place it along a hall symmetry line (to utilize the mirror function), at the stage wall, the stage front, or at the proscenium. If the hall has no typical stage, model the longest hall dimension as y since many plots are optimized for a longer y dimension. SI units are used but since the input data-files can use scale-factors for the coordinates of the hall model, it is possible to enter everything in any decimal unit. The output, however, always uses SI units. With the AutoCAD interface and DXF2GEO also imperial units can be used. Decimal numbers transferred to the CATT GEO-file format will be the same as inside AutoCAD since scale-factors are used to convert to meter when the files are read.

31

1. INTRODUCTION

Frequency range All calculations are, or can be, made for eight octave-bands: 125 to 16k Hz. In some cases calculation is also made for the linear or A-weighted power sum of these eight bands. If no data is available for 8k and 16k Hz, extrapolation is made from the values at 2k and 4k Hz. Note: the underlying theory is geometrical acoustics and in most halls the lowest two octaves will not be well predicted. For small rooms such as control rooms and studios typically only the upper octaves 1, 2, and 4k Hz will be well predicted (in all halls 8k and 16k Hz may suffer from lack of proper input data). The Interactive RT estimation dialog can give some info about a reasonable frequency range for a certain room model. For a discussion on the limitations refer to [Dalenbäck-95].

Source directivity Source directivity is either modeled by entering or importing horizontal and vertical polar values for every 15°, by importing measured data in a 10° full-space format, via the Common Loudspeaker Format [CLF], or via the DLL Directivity Interface (DDI) using the Directivity module. Directivity values can range from -50 dB to +15 dB relative the on-axis value. Number of directivity-files only limited by available disk space. Text import/export and conversion from DDI directivities to SD0 and SD1.

Data output Data output can be requested for any combination of up to 260 sources (A0 to Z9) and up to 100 receivers (00 to 99) and any combination of octave-bands plus the power sum of the octaves where relevant. Calculation types are indicated (G = Geometry view/check, Ad = Audience area mapping – direct only, Ad+r = Audience area mapping – direct + reflected, E = Early part detailed ISM, F = Full detailed calculation, D = Directivity module, S = Source addition module, P = Post-processing module, 3D = 3D-viewer): ·

echograms (ech.) (125 to 16k Hz + sum): ·

·

log scale (dB); ·

discrete reflections EF;

·

forward integrated ech. EF;

·

backward integrated ech. EF;

·

octave-band dependent integration to simulate the ear F;

·

directional X, Y, Z ech. F;

linear scale (Pa) E: ·

·

discrete impulses;

linear scale (Sone) E: ·

discrete impulses;

·

exponential octave-band dependent integration to simulate the ear;

·

“vectorgrams” F;

·

“sound roses” F;

·

EKgrad, echo disturbance F;

·

Spectra plots (direct, early, total) F;

32

1. INTRODUCTION

·

MATLAB results F;

·

impulse responses P;

·

transfer functions P;

·

image source space and number of impulses/second E;

·

history of reflections showing reflection trace, source and receiver angles E;

·

absorption coefficient histogram F;

·

free path histogram F;

·

plane hit histogram F;

·

room acoustic parameter overview for all calculated receiver locations F;

·

room acoustic parameter trend as a function of receiver number F;

·

RASTI/STI overview and trend FA;

·

direct sound level coverage (including directivity contours) A;

·

early sound level coverage in four user selectable time intervals A;

·

early lateral fractions in four user selectable time intervals A;

·

source directivity polar plots D;

·

source directivity 3D "balloons" D;

·

source directivity contours D;

·

3D projections of the hall (parallel, natural, and camera) G: ·

shaded 3D;

·

colored 3D;

·

shaded 3D for the standalone OpenGL-based CATT 3D-viewer 3D

·

reflection edge contours G;

·

source directivity -3 dB contours;

·

wire-frame with corner information;

·

wire-frame with interactive plane information (surface properties and areas, corners, plane normals);

·

plan, two sections, and a parallel 3D projection of the hall G;

·

plan, two sections, and source information G;

·

room acoustic parameters F: ·

Direct sound, SPL, Strength (G), RASTI, STI, Deutlichkeit (D-50/C-50), Clarity (C-80), Center of Gravity Time (Ts), Lateral Energy Fraction (LF,LFC), EDT, Reverberation Time (Sabine, Eyring, T-15, T-30), Stage Support, U50, AI, PI, SII, IACC etc.;

·

color mapping of most parameters, also source delays and source group with shortest delay can be displayed Ad Adr.

·

spreadsheet export for e.g. MS Excel F Ad Adr.

·

room acoustic parameters can optionally be displayed normalized with expected values, according to a theory developed by [Barron-88] (see Prediction module Theory) F;

·

geometry debug-file (geometry check, plane normals, which planes uses a certain corner/node, reversed planes etc.) G;

·

geometry tree-file (the order in which GEO-files are included) (Prediction module);

·

geometry/absorption statistics (surface used by each material etc.), classical RT G;

33

1. INTRODUCTION

·

geometry export-files for the AutoCAD interface that enables files made directly in the CATT format to be imported into AutoCAD (Prediction module);

·

geometry export-files for VRML 2.0 browsers that enable interactive 3D web-sites including auralized sound (Prediction module).

·

geometry export-files to DXF.

·

geometry export as CATT marker-files (for use as non-acoustical 3D elements or loudspeaker visuals) G

1.6 How to get started For the first project it might not seem worth while to use the advanced possibilities of creating a structured geometry and use named constants and expressions to define corner/node coordinates etc. This is probably true if one gets everything right the first time and if the hall design never has to be changed or fine-tuned. Most projects, however, require a lot of finetuning of the design. Often some of these changes can be anticipated and be allowed for by using named constants (e.g. reflector heights or angles). The extra time spent planning in the first part of the project will almost certainly pay off in the later part. If only numbers are used it is very easy to paint oneself into a corner. For the first project it is recommended to create a very simple hall, such as described in Section 1.2: ·

create a rectangular (shoebox-shaped) hall and use named constants for width, depth and height

·

add a simple rectangular reflector in a separate file and include it. Define the reflector-file as an object-file and try translate, rotate and copy it.

By experimenting with these two simple parts (changing constant values and reflector rotation etc.) the basic knowledge on how to model a hall in CATT-Acoustic is acquired. A more complex hall takes more time and more planning but the procedure is the same.

1.7 About CATT A short presentation of me - Bengt-Inge Dalenbäck, the program developer - and the company CATT. I acquired an M.Sc. in electrical engineering 1980 with a thesis in semiconductor technology, working on a gas sensitive Field Effect Transistor (FET). I have been involved in acoustics since 1981, the years up to 1989-90 mostly as a part-time teacher at Dept. of Applied Acoustics at Chalmers University of Technology, Gothenburg, Sweden, see [Chalmers]. 1986 I started an independent company CATT (Computer Aided Theatre Technique) originally developing theater lighting and decor design CAD software and was until 1996 half-time occupied at the department and half-time at the company. Since 1988 CATT has concentrated on software for room acoustics. 1990, I joined the Chalmers Room Acoustics Group for half-time Ph.D. studies [Chalmers]. See the www page for a list of publications. 1995, December, I received a Ph.D. on room acoustic prediction and auralization at Chalmers [Dalenbäck-95b]. A list of main programming projects and developments within CATT:

34

1. INTRODUCTION

·

1987, CATT-Lighting: 3D CAD-program for theater lighting design (built on ray-tracing).

·

1987, CATT-Decor: 3D CAD-program for decor design on theater stages aimed to support CATT-Lighting. With an object library and with hidden line removal.

·

1988, CATT-Acoustic: Room acoustic simulation using the image source model. The first three projects were implemented on a Commodore Amiga.

·

1989, CATT-Acoustic MS-DOS. The lighting and decor design programs are dropped and the work concentrates on room acoustics and PC-software.

·

1990, Binaural post-processing of the results from CATT-Acoustic creating binaural impulse responses to use for audible simulations (so-called auralization). For convolution a Lake FDP-1 convolution processor was required [Lake].

·

1990-1994, Various enhancements and upgrades of CATT-Acoustic.

·

1994, Software convolution enabling auralization using only a PC sound-card. Specialized hardware is now optional.

·

1996, June, release of CATT-Acoustic v6 for Windows 3.1.

·

1996, July, goes from a 50% to a 10% position at Chalmers fully concentrating on CATT.

·

1996, August, release of CATT-VRoom designed for full frequency-range studio and virtual reality reverb mainly together with a Lake DSP Huron or CP4 convolution processor.

·

1997, March, CATT-Acoustic v6.1 with audience area color mapping.

·

1998, February, release of CATT-Acoustic v7.0 for 32-bit Windows introducing the RTC.

·

1998, May, addition of a loudspeaker replay option (cross-talk cancellation)

·

1998, October, v7.1 introducing DDI and array modeling.

·

1999, October, v7.2 introducing RTC-II and the OpenGL-based CATT 3D-viewer.

·

2000, The FIReverb Suite, for natural music FIR reverb and multi-channel convolution (CATT-VRoom is enhanced, renamed to PureVerb and supplemented by the MultiVolver).

·

2001, The FIReverb Suite 2 Ed., 24- and 32-bit processing, Ambisonic decoder.

·

2002, February, CATT-Acoustic v8.0 with 8k and 16k Hz support, material colors, visual markers, walkthrough convolution, 5-channel post-processing.

·

2002-2005, development of the Common Loudspeaker Format [CLF] together with other prediction software developers, ETC, and the SynAudCon.

·

2003, The FIReverb Suite 3 Ed., graphical selection of room and 5-channel mic setups.

⋅

2003, February, frequency dependent semi-transparency, ray-, echogram- and mapanimation.

·

2004, January, 2 order B-format impulse responses.

·

2004, March, pixel rendering of direct sound.

·

2005, March, interactive RT estimate using global ray-tracing, GratisVolver free convolver for WAV-file format impulse responses.

·

2005, May, CATT-Walker real time walkthrough auralization, support for the Common Loudspeaker Format [CLF]

·

2006, January, separate direct sound and direct+reflected sound audience are maps, 3D audience area maps.

·

2006, October, noise mapping, mapping speedup, 3 order B-format impulse responses, Walker Steer SDK for CATT-Walker, GratisWavPlayer a small free player for the WAV-file format.

·

2007, April, MATLAB/EXE and script interface for Sequence processing (eg. for parameter optimization), updated and improved Help in Compile HTML format, mouse plane picking

nd

rd

nd

rd

35

1. INTRODUCTION

in PLT and OGL-viewers. Separation of programs and other read-only files and data libraries (CATTDATA). ⋅ ⋅

2007, August, optional merging of planes to larger polygons, new printed manual. CATT can be reached at:

CATT Mariagatan 16A SE-41471 Gothenburg SWEDEN

Phone/fax: +46 31 145154 e-mail: [email protected] www: http://www.catt.se

36

2. PREDICTION MODULE

2. PREDICTION MODULE The Prediction module is the core of CATT-Acoustic and offers three main prediction types: early reflections based on the Image Source Model with added first-order diffuse reflection, audience area color mapping (direct sound and direct+reflected sound), and full detailed calculation that exports data for further processing in the Source addition and Post-processing modules. In addition direct and reflected sound pixel rendering and interactive ray-tracing based reverberation time estimates can be performed. The Prediction module uses settings-files with the extension .PRD that stores all necessary information to perform a prediction. Several different settings-files can be created for different purposes. Settings-files can be used in the Sequence processing module for batch processing, which is an important feature since room acoustic prediction typically requires long calculation times. Output/result-files always have the same names independent of actual project. It is only the output folder that is project-specific. Output-files created by the various settings have the source id, the receiver id, and the octave-band coded in the file-names as: ss

source id

[A0,A1,...,Z8,Z9]

rr

receiver id

[00,01,...,98,99]

oct

octave-band

[ 125 250 500 1K 2K 4K 8K 16K SUM ]

Full detailed echogram plot-file names, for example, have the syntax E_ss_rr_oct.PLT. The filename for source B2 and receiver no. 13 and octave-band 1 kHz will thus be E_B2_13_1K.PLT. An overview of all output-files is presented in Section 1.4 and Section 2.4 explains each prediction output-file in detail.

2.1 Menus File|New Settings...

Create a new prediction settings-file. If a Template.PRD file exists in the CATTDATA-folder it will be used as a template for the created settings-file. File|Open Settings...

Load a new prediction settings-file (.PRD). File|Save Settings

Save current prediction settings. File|Save Settings As...

Save current prediction settings under a new name.

37


File|Export Geometry To|AutoCAD Interface CAD...

Exports to the CATT AutoLISP AutoCAD interface, see Section 10. Current General settings are used to create the file (input folder, output folder, master-file, plane use, source-file, receiver-file, used sources and receivers). The CAD-file always uses decimal point and can also be used for exporting the geometry to other similar software. Note: only coordinates actually used in the model are exported. At export, an option is presented to also include a list of the surface properties used, this is not to be used for exporting to the AutoCAD interface but for other prediction software such as ODEON. If selected, a section is created with the header %SURFACEPROPERTIES and each property used is listed with the following syntax: propname

a125 ... a16k [ diftype s125 ... s16k ] [ SEMITRANSP t125 ... t16k ] {R G B}

where: diftype is L

items in brackets [] are included only if the data type is used. ...

means a sequence of coefficients from 125 to 16k Hz (given in % with one decimal).

R G B means the Red, Green and Blue surface color values [0..255] File|Export Geometry To|VRML Browser WRL...

Exports to the VRML 2.0 .WRL format that can be used by VRML 2.0 browsers/plug-ins. VRML stands for Virtual Reality Modeling Language and is a standard file-format for virtual 3D worlds (WRL is short for world). Once loaded into a VRML browser/viewer the geometry can be manipulated directly (changing view point, walking around etc.). Each plane is saved as a primitive to allow for individual coloring in 3D-modeling software e.g. 3ds max (any spaces and other illegal VRML node name characters are replaced by "_").

38

2. PREDICTION MODULE Show source and receiver ids

VRML export

Includes “01”, “A0” etc. as 3D text above each receiver (sphere) or source (box). Show edges

VRML export

Outlines all surfaces with visible edges. Experiment to see which setting works best for a particular model. {

Add comments

VRML export

Adds comprehensive commenting to the VRML structure to simplify manual editing. Since the comments take unnecessary space only switch on when needed. Add sound templates

VRML export

Includes a template for playing sound associated with each source and receiver. The template can then be edited in the WRL-file and a WAV-file name be added, see below. Automatic lighting

VRML export

Includes ceiling lights where the fixtures are invisible. The intensity is calculated to give a good lighting in models of all sizes and shapes. If the model still becomes too bright or too dark, the Pointlight node towards the top of the file can be edited (to decrease the light decrease the intensity, to increase the light decrease the non-zero attenuation field. If no lighting is included the exported model needs to be seen with the Headlight switched on in the viewer or lights have to be added in a VRML editor. Triangularize planes

VRML export

Normally the VRML browser performs the triangularization but it has been discovered that some VRML editing software (e.g. VRCreator) cannot handle faces with more than four corners. Indent node structure

VRML export

Indents each successive level in the VRML structure to simplify manual editing. Since the indentation takes unnecessary space only switch on when needed. Background (Colors)

VRML export

The background "sky" color outside the model. Simple (Colors)

VRML export

Use one color for audience surfaces and one for remaining surfaces. Audience (Colors)

VRML export

For the Simple case: the basic color of audience surfaces (as selected in Audience area mapping). Surface (Colors)

VRML export

For the Simple case: the basic color of a non-audience surfaces.

39

2. PREDICTION MODULE As defined (Colors)

VRML export

Uses the color assigned to each surface property. Basic browser operation: ·

to view locally using an HTML browser Open the WRL-file and, if installed, the browser will start a VRML browser plug-in that can view WRL-files. VRML browsers sometimes come included with HTML browsers or can otherwise be downloaded from various sources. Good places to look is www.w3.org/MarkUp/VRML and www.web3d.org. A WRL-file can also be added to a home page and be viewed over the network by WRL-aware browsers (either use very short sound-samples at 22.05 kHz or compressed file-formats for this or down-load times will be long).

·

switch off "gravity" and "collision detection" since these are appropriate mostly for walkthrough etc.

·

each receiver position has its own "Viewpoint" set with the head direction used. A list of Viewpoints is often placed on a browser menu (normal and/or right mouse button menu). The source gets its own Viewpoint with the direction of the source aim. If the mouse is placed over a receiver or source the cursor turns into a hand or similar and if clicked and held down the view temporarily becomes that of the receiver or source. To have the view stick use the Viewpoints menu.

·

auralized sound can be played (if Add sound templates is checked when exporting) by changing the "url" of each receiver (say 01) from "" to an actually created WAV-file e.g. "file://c:\wavs\MT_A0_01.wav" i.e. auralized sound for receiver 01 (WRL-files are plain text-files and can be loaded into an editor: search for "url"). When the Viewpoint is selected a small white sphere can be seen at the center of the screen (it is placed directly in the head direction) and if it is clicked the WAV-file is played. Also the source can have a WAV-file in a similar manner and if the anechoic WAV-file is used it can serve as a reference to the auralized sound. Remember to perform a relative calibration on all auralized WAV-files included.

Note: many of the features of the VRML export viewing, plus more features, were later made available in the CATT 3D-viewer and in a more straightforward and simple to use way. File|Export Geometry To|CATT Marker MRK...

Exports to a CATT marker-file that then can be loaded as an object in a GEO-file MARKERS section using LOAD or be used as source directivity visuals. For pure geometry export as a marker it is most convenient to add the directive NOROOM into the master-file directives section. With a NOROOM GEO-file the source and receiver files do not have to be created. An option will be presented to export a combination of planes, plane edges and markers. If planes are not exported they can be used as reference planes for exported marker objects (e.g. a floor) while building the geometry. Sub-divided planes are exported as a single plane with the color of the first sub-division (this export function is intended mainly for markers). File|Export Geometry To|CLF Cabinet Geometry...

Similar to when exporting to a CATT marker-file for source visuals, a geometry can be exported as an edge and face list for a CLF text-file geometry [CLF] that can be pasted into a CLF text-file. This option is of course useful mainly to CLF authors and a benefit is that the GEO format allows for parametrical modeling of cabinets and allows e.g. for re-use of similar cabinets with changed measures. File|Export Geometry To|DXF...

Similar to when exporting to the AutoCAD Interface CAD-format, a geometry can be exported to DXF. This option is of useful for export to other prediction software or to CAD programs. The export uses only 3DFACEs, so planes will > 4 corners will be triangulated before export,

40


sources, source visuals, receivers and markers are not exported. The surface properties are color-coded using AutoCAD 14 palette and color indices and an accompanying text-file lists the properties for each material/color (name.DXF gives a text file name_dxf_surface_properties.TXT). Note: sub-divisions are exported as separate 3DFACEs so if overlapping subdivisions have been used (such as a door and a window in a full wall) the importing program will have to adjust the handling of such surfaces according to the method it uses. There may be other changes necessary in the geometry to make it suitable for the importing program but the bulk of the work should be done. File|Export Geometry To|Merged Planes GEO-file...

Exports to a new CATT GEO-file were planes have been merged to larger polygons as far as possible. This option is useful for GEO-files created via the AutoCAD interface or converted from DXF via DXF2GEO, see section 10. All planes are saved into a single selected GEO-file where planes that have been merged are commented out by ;merged and new merged planes are named merged. Planes that have a near identical plane equation and the same absorption property are merged so it can be useful to assign properties before merging. The merged geometry may have a few planes that have errors, or that could not be merged, and need to be treated manually but the bulk of the work should be done. After a successful merge, the new file can be INCLUDEd and a RETURN statement added at the top of the original file’s PLANES section. Note: sub-divided planes are never merged (files from the AutoCAD or interface or converted via DXF have no sub-divided planes). Utilities|Latest Results List

Each time text-based result-files are created the corresponding filenames appear in a small list in the main window. This item toggles the list on and off while preserving its contents. To load a file for viewing in CATT-Edit double-click the filename. Utilities|Create a New Project...

Displays a dialog prompting for a project name and folder.

Template GEO- and LOC-files are then created ready to be filled in (a sub-folder OUT will be created to be used as output folder). Also a basic settings-file GEO.PRD and a text-file PROJECT.TXT (optionally to be used for a project log) is created, see Section 11, Tutorial. Utilities|Create a GEO-file...

Creates a GEO-file template with selected name. CATT-Edit can also create a GEO-file template. Utilities|Copy a GEO-file...

Copies a GEO-file, e.g. from another project where a special kind of reflector was created, to the project folder under selected name.

41

2. PREDICTION MODULE Utilities|Backup Project...

Displays a dialog prompting for the backup folder to use and which file-types that are to be included in the backup.

It is also a good idea to regularly backup the source directivity files in the CATTDATA-folder SD. Utilities|Create GEO-file Tree

Creates a text-file displaying all included geometry-files (in the order they were included), the source- and the receiver-file together with other information useful when planning or debugging complex projects. Current General settings are used to create the file (input folder, master-file, source-file and receiver-file). The file created has the same name as the masterfile but _tree.TXT is added. The file is immediately after creation loaded into CATT-Edit in view-only mode.

42


2.2 Dialogs

The main Prediction dialog controls settings for the combination of actions to be performed and has an interface to CATT-Edit. When Save and Run is clicked the current PRD settingsfile is saved and the Prediction module performs the actions requested. If warnings are on (Preferences), the output folder is checked for result-files that may be overwritten. Independent of Geometry view/check debug settings, as soon as any type of prediction is requested a check for duplicate plane and corner ids is performed and if found prediction will not be possible. The actions checked are independent but it can be useful to always include the Geometry view/check so that the latest geometry plot-files correspond to the latest results. The two items Interactive RT estimation and Direct sound pixel rendering are special functions that will use the General settings options but do not affect the settings-file.

43


General settings The General settings dialog controls where to read and write data and which files to use as well as other global settings e.g. for diffuse reflection.

Project

General settings

A project name for reference. Only the first 70 characters are used. The string will appear on all plots (the first 40 characters) and in all result-files. Input folder

General settings

The project files are always read from this folder: geometry-, source-, and receiver-files. A dot (.) can be used as shortcut for the folder where the PRD settings-file (at prediction-time) is located so that a complete project can be moved without changes to the PRD. Output folder

General settings

In this folder all output-files are created, if it does not exist it will be created. To compare two calculations change output to a new folder (typically a sub-folder to the Input folder): Input folder Output folder, hall version 1 Output folder, hall version 2 A dot (.) can be used as shortcut for the folder where the PRD settings-file (at prediction-time) is located so that a complete project can be moved without changes to the PRD. e.g. .\OUT. If the output folder does not exist at prediction-time it will be created.

44

2. PREDICTION MODULE Master-file

General settings

Which geometry-file (in the Input folder) to use as a master-file (i.e. root of the geometry-file tree). The master-file is the file that is read first and may then INCLUDE other files (in the Input folder) that, in turn, may include further files etc. By having more than one master-file e.g. including different geometry parts, a geometry version can be changed just by selecting a different master-file. The master-file is usually named MASTER.GEO (GEO is mandatory extension). When (Edit) Geo-file on the main dialog is clicked the specified master-file is initially selected in the Open file dialog. File (Sound sources)

General settings

Selection of the source-file that contains source locations, aims, directivities, delays, source roll and octave-band levels and more. By having more than one source-file e.g. including different directivities or aims, a new setup can be tested by selecting a different source-file. The source-file is usually named SRC.LOC (LOC is mandatory extension). When (Edit) Source-file on the main dialog is selected the specified source-file is initially selected in the Open file dialog. Which to use (Sound sources)

General settings

Which of the sources defined in the source-file to use.

If source ids are chosen that do not exist in the source-file it is not trapped until the prediction is actually run. Show all

General settings/Sources

Displays all selected sources. (Group) On / Off

General settings/Sources

Selects / deselects all sources in a group. File (Receivers)

General settings

Contains receiver locations and optionally individual head directions. By having more than one receiver-file including different sets of receiver positions, a new setup can be tested by selecting a different receiver-file. The receiver-file is usually named REC.LOC (LOC is mandatory extension). When (Edit)Receiver-file on the main dialog is selected the specified receiver-file is initially selected in the Open file dialog Which to use (Receivers)

General settings

Which subset of the positions defined in the receiver-file to use.

45


(Group) buttons affect the receivers in view, (All) buttons affect all receivers. If positions are chosen that do not exist in the receiver-file, it is not trapped until the prediction is actually run. (Plane use)

General settings

Options to use only a subset of planes defined in the geometry-files. A dialog is displayed allowing for ranges of plane ids to be included or excluded (to select a single plane enter the same id in both edit boxes or leave the to box empty). The Free form edit button enables giving the plane ids "by hand" instead, e.g. 1 4-9 15 33-40. Any ranges already in the list-box will be converted to a string and after editing they will again be listed in the list-box. This is very useful e.g. for making a clearer view while creating a complex and/or mirrored model (exclude the mirrored planes) or to test with/without a particular group of planes e.g. reflectors. Note: the full geometry is always read and planes not to be used are excluded after the GEO-files have been read.

(Head direction)

General settings

The head direction is required for lateral energy fraction measures and for post-processing using the Post-processing module. The settings below apply for all receivers that do not have an individual head-direction as selected in the receiver-file. Source

Stage

looking towards the source. Note that this setting leads to varying head-directions when multiple sources are used. In such a case - set headdirection Fixed pos to the main source, i.e. the source to be considered as the reference (e.g. the speaker). turned in the direction of the negative y-axis - all receivers (listeners) will have "stiff necks" looking towards the stage. Note the coordinate system to be used (Sect. 1.5).

22 A0 Stage

12

22 A0 12

Fixed pos heads turned to the coordinate defined.

22 A0 12

46


Acoustic environment

General settings

Displays a dialog for air absorption, air temperature, air density and background noise levels. Temperature

General settings/Acoustic environment

Affects speed of sound, c, and estimated air absorption, see Theory. Humidity


Affects estimated air absorption, see Theory. Density


Air density, ρ, affects wave impedance, ρc. (Air absorption)


Off

no air absorption applied. This option is for special purposes.

Estimate

air absorption is estimated from air humidity and temperature. See Theory.

User defined

the edit button displays a dialog with octave-band scrollbars. These scrollbars all have different ranges (air absorption is very frequency dependent). The parameter that is selected is the energy dissipation coefficient, mb, see Theory.

Background noise


Displays a dialog with octave-band scroll-bars. Note that background level is actually used only for STI (included bands, see Preferences), RASTI (only 500 and 2k Hz values) and U50 (available via hidden option, see 1.3). The NCB curve that corresponds to the levels given in: [ANSI S12.2-1995], Table 1, Balanced noise criterion (NCB) curves. The indicated dBA value includes the 16k octave. Without a noise map / Total average is used as a constant background over the room. With a noise map Residual average is used as a constant residual background over the room that is power added to individual receiver noise in Full detailed calculation or to a noise map in Audience area mapping. Tip: to set a certain NCB value start with low values on all sliders and raise each band until it goes up to the curve desired. Note: NCBs are actually curves between 31.5-8k Hz but only the range 125-8k Hz is used. (Diffuse reflection)

General settings

If and what type of diffuse reflection to use.

47

2. PREDICTION MODULE Off

no diffuse reflection at all. This option is for special purposes e.g. to estimate a given geometry's sensitivity to diffuse reflection but never for normal calculations.

Surface

surface diffusion according to scattering coefficients assigned to each surface property or via Surface default scattering coefficients.

Surface + edge

also includes automatic size and frequency dependent "edge diffusion". Note: the automatic diffusion calculated from edge area/wavelength ratio only takes effect for surfaces that have an "*" appended to their material property (typically reflectors), see Section 2.3 and 2.5.

Surface default

default scattering coefficients that will be assigned to all planes that have not been given specific scattering coefficients. This should normally always be selected and set to 10% corresponding to flat surfaces that are not small. See Calculation recommendations, Section 2.5.

Geometry view/check The Geometry view/check dialog controls settings for checking geometry errors, consistency, and generation of geometry plot- and debug-files. For each plot-file type and Shaded.OGL any non-acoustical visual markers can optionally be included via the * checkboxes.

Four views (2D) (Plot-files)

Geometry view/check

Four views of the geometry, VIEW4.PLT. Plan, two sections and an auto-scaled parallel wire-

48


frame 3D projection of the hall. If audience planes are selected in Audience area mapping they are marked gray in the plot and the total audience area is displayed. The volume is calculated and displayed (this is only meaningful for closed rooms). If only a geometry check is performed the volume calculation is less accurate to save time. Source info (2D) (Plot-files)

Geometry view/check

Similar to VIEW4.PLT but the lower right pane gives information of each source selected (SRCINFO.PLT). For each source one line is shown with a Up ring for the source aim (where "front" or "on-axis" is in the directivity definition) and another line is shown with an arrow defining the "up" direction according to the optional source "roll" angle (where "up" is in the directivity definition). From these "left"/"right", "back" and "down" can easily be worked out. Shaded (3D)(Plot-files)

Front

Geometry view/check

A 3D-projection with shaded surfaces (SHADED.PLT). The shading is viewing-angle dependent and is made on the reflecting side of each surface thus making it easy to see if a plane has been reversed or if the model has holes. Colored (3D)(Plot-files)

Geometry view/check

A 3D-projection with colored surfaces (COLORED.PLT). The colors are those assigned to, or auto-generated, for each surface property. Source info (3D) (Plot-files)

Geometry view/check

A wire-frame 3D-projection where the -3dB contours for selected bands are shown in addition to source-file data (SRCINFO3D.PLT). The contours of not selected sources are shown gray. For details on directivity contours, see Section 4.3. Plane info (3D) (Plot-files)

Geometry view/check

A wire-frame 3D-projection with interactive plane, plane corner and other information, PLINFO.PLT. When stepping through the list of planes their front (reflecting) side should be colored with the assigned (or auto-generated) color and the back side be black. If the color appears on the wrong side the plane corner entering order must be reversed. This file is the major tool for checking the geometry built in the geometry-files. Corner info (3D) (Plot-files)

Geometry view/check

A wire-frame 3D-projection with corner id information, COINFO.PLT. Refl. info (3D) (Plot-files)

Geometry view/check

A wire-frame 3D-projection with interactive reflector edge contours of selected planes, plus plane and plane corner information, REFLINFO.PLT. Planes are colored with the assigned (or auto-generated) color on the reflecting side and black on the back side. This is a purely geometrical construct and any effects of diffusion and/or diffraction has to be considered separately. Note that no shadows are cast, only reflected edge points are calculated. If more than one source is used, each plot frame shows reflections from both sources. The edge contours of not selected planes are shown grey. (Primary view)(3D)

Geometry view/check

The primary view controls which of the 3D-projection methods to use as the primary view. The 3D-projection method can always be changed in the View module but it is practical if the

49


default setting is appropriate for the particular model. (Initial view parameters) (3D)

Geometry view/check

The 3D-view for 3D and 3DA files can always be changed in the View module but it is practical if the default settings are appropriate for the particular model. Once a good view has been selected in the View module it can be copied to the prediction settings using Edit|Copy 3Dview to PRD. View distance

Geometry view/check/View settings

View distance from the room coordinate system origin. Lens angle

y view/check/View settings

Lens angle (field of view) for Natural Horizontal view angle

y view/check/View settings

-180° to 180° where 0° is a view along the y-axis. A positive value rotates the view position to the right and a negative value to the left.

Vertical view angle


-90° to 90° where 0° is in the horizontal plane. A positive value gives a view from above and negative value from below. Horizontal (X) shift


Horizontal translation of the screen image (does not move the view position but only the screen image). Unit is generic pixels where the full plot window width is 640. Positive values move the picture to the right while a value of 0 places the rotation center at the window's horizontal center. Vertical (Y) shift


Vertical translation of the screen image (does not move the view position but only the screen image). Unit is generic pixels where the full plot window height is 385. Positive values move the picture down while a value of 0 places the rotation center at the window's vertical center. (Position)

Geometry view/check/Camera settings

Coordinate of camera position. (Target)

iew/check/Camera settings

Coordinate of camera target. Lens angle

view/check/Camera settings

Camera lens angle (field of view).

50


Shaded.OGL (For 3D-Viewer)

Geometry view/check

A file similar to COLORED.PLT is created, SHADED.OGL, for the stand-alone CATT 3D-viewer (CATTOGL32.EXE) with extended 3D viewing possibilities where also a set of planes can be selected and used in the various plane selection dialogs such as for Audience area mapping Audience planes. The OGL-file contains all source and receiver locations as viewpoints. The viewer is based on OpenGL and is documented separately by its help-file (CATTOGL32.CHM). Auto load file in viewer (For 3D-Viewer)

Geometry view/check

Automatically loads the created file in the viewer (or updates it if it is already loaded). Include contours (For 3D-Viewer)

Geometry view/check

Includes -3, -6 and -9 dB source directivity contours for aiming help in the viewer when using camera mode and a source viewpoint. For details on directivity contours, see Section 4.3. (Debug)

Geometry view/check

What type of geometry debug to perform while processing the geometry. Immediate (Debug)

Geometry view/check

Checks for major errors and terminates immediately with an error message. The same checks as for Create file are performed but the program terminates immediately if an error is found. The "inaccurate corner check" checks for large errors thus will only detect big mistakes. It is checked that sources and receivers are inside the room. Just like for volume and reversed normal calculations, this test relies on that the surfaces are modeled correctly. If an open room is already detected (because an actual open room or due to modeling errors) the test is not performed. It is not necessary to perform this check once the basic model is ok. Create file (Debug)

Geometry view/check

Creates a text-file DEBUG.TXT that lists all errors found and documents the geometry.

51


The checks optionally performed are: ·

check of duplicate corner ids (must be corrected, will block all prediction)

·

check of duplicate corners (same or nearly same x, y, z but different ids).

·

check of duplicate plane ids (must be corrected, will block all prediction)

·

check of duplicate planes (same corners and same normal, must be corrected). Doublesided surfaces like reflectors use the same plane corners but have opposite normals.

·

check of inaccurate corners (within a certain accuracy). Note: it is not possible to know which corner caused the error. If the plane equation is built from one inaccurate corner and two accurate corners, a fourth accurate corner will seem to be out of the plane. Must be corrected or rays will be lost and an open room may be declared.

·

check if planes are coinciding/overlapping (i.e. e.g. a carpet modeled on the floor, use subdivisions instead). Overlapping planes will cause volume and mean absorption errors and "open room".

·

check if plane edges are penetrating/touching other planes (e.g. a reflector through a wall). Penetrating planes will confuse the pre-processing of the geometry to speed up ray-tracing and must be corrected. Touching planes are not critical and is often the result of an AutoCAD export.

·

check if corners are used in only one plane ("single-connected corners"). This should never happen with closed rooms;

·

check if some planes may be reversed. This check only works for closed rooms and is not 100% reliable since it depends on how the model has been built but serves as a check for potentially reversed planes. The major tools for checking are the PLINFO.PLT, SHADED.PLT, COLORED.PLT plot-files and SHADED.OGL;

·

a list of all corners used and in which planes they are used (sub-division corner ids are not listed);

·

a map of all plane ids used

·

a list of all planes used (sorted by plane id) with a detailed description of each plane: id, name, type, plane equation, absorption/diffusion properties, sub-divisions, corner ids.

52


Writing a massive amount of text information to disk takes time. Especially the option for listing all planes can be switched off to make the DEBUG.TXT file smaller. Apart from the list of planes option, all options should normally be on but if a lot of errors of one kind are found switching off some options temporarily will make the debug file faster to write to disk and simpler to read. Create .CMB without calculation

Geometry view/check

If a source combination (.CMB) file should be created. This can be useful when working with the Source addition module since geometry plot-files can be created that display source aims and source combination and at the same time saves a corresponding combination-file that the Source addition module can read. Filename Name.CMB if Name.LOC is the selected sourcefile name under General settings. Note: if source addition is selected in Full detailed calculation or Early part detailed ISM dialogs a CMB-file is always generated for the current source-file. Assume closed model for Sabine RT

Geometry view/check

If a model should be considered as closed even if the automatic detection finds it open. Even if a model looks closed it can be found to be open if the geometry contains too many errors or is built too inconsistently (overlapping surfaces, reflectors that are not double-sided, warped planes, see 2.3). By checking this option a volume estimate will still be given in VIEW4.PLT but enclosed in parenthesis. With a Full detailed calculation, a model is declared open if too many rays are lost and is not affected by this setting.

Audience area mapping The Audience area mapping dialogs control the estimation of early SPL and LF and most parameters (D-50, C-80, RASTI etc.) over a grid covering selected audience-planes. The mapping is available in two versions: direct sound only (requiring no rays) and direct + reflected requiring ray-tracing. In both cases source directivity and multiple sources with delays are taken into account (delays are set in the source-file). The two types of mapping share the same Audience planes so if one is changed the other follows.

53


54


Direct + reflected only: Number of rays/oct (Settings)

Audience area mapping (d + r)

Number of rays to use for coverage mapping. At least 10000 rays must be used for reasonable mapping. In general, the smaller the Map step the higher the number of rays has to be. With low ray-numbers the results will be too random to be useful. See also Auto number. Note: to decrease calculation times for big projects where some sources are considered main sources while others may just be fill sources, the latter can use less rays. To enable this, a MAPRAYFACTOR may be given for each source in the source-file. If the factor given will make the number of rays used be less than 100 it will be selected as 100 (no matter the number of rays the direct sound is the same). If this option is used it is documented in the plot-file title bar as /F and is also indicated in MAPSTAT.TXT along each source listed Auto number (Settings)


Automatic selection of a good/sufficient number of rays. This number is dependent on the room size, the map step and some other parameters. 1) One number is calculated to give one 2 direct ray / m at a time corresponding to 80 ms arrival relative the direct sound for all positions, this gives a good sampling of parameters such as D-50, C-80 and LF. 2) One 2 number is calculated corresponding to one direct ray / 4 m at the longest hall dimension. 3) One number is calculated corresponding to one ray / map point at 80 ms relative arrival time. The highest of 1), 2) and 3) is selected. 2) will only be higher for rooms that are very big so that there is some chance to get back some detailed reflections from far-away surfaces and 3) will only be higher for very dense maps. Note: if a MAPRAYFACTOR is used, see above, it is applied to the auto number. Adapt rays (Settings)


This option will speed up mapping with a factor 2-20x typical gain is 3-4x. The adaptation is th th made by decimating the number of rays used after the 4 reflection order and for every 4 subsequent order after that. This means that the early reflections are handled the same way and e.g. strong low order echoes are included, and only the later reflections where, for measure-calculation, mainly the total energy matters. After testing many rooms the speed-up seems very safe but it is still advised to use it only on projects where calculation time is a real problem. If this option is on it is documented in the plot-file title bar as /A and is also indicated in MAPSTAT.TXT. Ray truncation-time (Settings)


Truncation time in ms for rays to be traced. The time should correspond to at least 50% of the reverberation time (measures such as D-50 etc. are not much affected by the energy 30 dB down from the direct sound). See also Auto time. Auto time (Settings)


Automatic selection of sufficiently long truncation time based on the estimated classical RT (only possible if rooms are closed and if classical RT does not differ too much from the actual one e.g. because of double-sloped decays).

55


Animate rays (Settings)


Makes it possible to see the actual rays traced and the emerging SPL map: ·

with animation on, the calculation time increases considerably but when switched off (Animate) No nearly the normal speed is resumed. If no animation at all is selected the speed difference from previous versions is negligible.

·

with Sequence module batch predictions, no animation is shown independent of selections.

·

the 3D model can be rotated or zoomed at any time (also while animating) using the same mouse commands as in the PLT-viewer and the 3D-viewer (LMB rotates, Ctrl-LMB or LMB+RMB shifts, RMB zooms in/out). The rotation is a bit more responsive when Step is selected but 3D manipulation can be done while calculating.

·

the 3D model always uses a parallel perspective.

·

the 3D view of the hall assumes that the coordinate system used has its origin inside or close to the actual hall geometry or it can be hard to find the model (use Ctrl-LMB).

·

when rays are stepped one at a time, the current pictures can be copied and pasted into e.g. Word. Use (Copy) 3D for the 3D view of the room and (Copy) 2D for the current echogram or map.

·

for each source the rays are color-coded Red-Yellow-Green-Magenta-Blue-Black-Light grey where each color transition covers 10 dB (i.e. 0-10-20-30-40-50-60 dB, below 60 dB the ray is truncated in the animation). Pure red is the first ray segment assuming an omni source and if the source is directive the directivity is applied first (the color of a ray from the rear of a very directive source may thus start out e.g. blue but on-axis always red, see about color-code for multiple-source mapping below).

·

direct and 1 order rays are drawn with heavy lines to make it easier to follow the rays.

·

if an animation is stopped (No) the calculation continues and if the animation is resumed (Yes or Step) the map is updated to include all previous hits.

·

any of the bands selected for calculation can be animated.

·

the emerging map represents the total relative SPL where Red corresponds to the strongest direct sound at a map point.

·

each ray hit at a map point is highlighted with the current relative ray level.

·

when direct sound is added with phase the direct sound component is the last to be added to the map, while with power addition it is the first.

st

Noise mapping (Settings)


As an alternative to using the constant background noise in General settings/Acoustic Environment (Env.) in STI and RASTI calculations, a predicted noise map can be used so that at each map point the predicted noise SPL is used rather than a constant over the room.

56


The procedure is as follows: ·

select all sources, noise as well as speech in General settings

·

if a noise map is used, a new residual background noise in Env. can be added to the noise sources, it could be e.g. caused by external noise or a well distributed ventilation system.

·

in the mapping dialog, check Noise mapping and select which of the sources are to be considered as noise sources (it may be all of them, see note below).

·

run the prediction. In a first phase a separate mapping only for the noise sources will be made creating a noise map (SPL only, PLT-files are named NOISEMAP_oct.PLT and NOISEMAP_oct_3D.PLT) and directly after a second phase follows using only the speech sources. When STI and/or RASTI are finally calculated the noise map is used for the background noise. A special MAPSTAT_NOISEMAP.TXT is created.

·

in STI and RASTI plot-files the fact that a noise map is used instead of the overall background noise is indicated .

Note: the noise map prediction will use the selected ray number/truncation time and map step/height settings but any selected Adding sound with phase and Animate rays will be switched off during the noise mapping phase (no point with noise sources to include phase). Often a high density map and many rays are wasted for a noise map but the noise sources can use a MAPRAYFACTOR < 1.0 to make that part of the calculation faster. It was decided to force a calculation of the noise and the speech at the same time, rather than to allow a noise map to be first calculated and stored, since it would then be complicated to ensure that the two cases are compatible. In some cases, the result of a noise map may be found constant enough and be entered as a new constant background in Env. and subsequent STI/RASTI mappings need not use a noise map. Note: if a pure all-noise SPL estimation case is of interest, such as with factory noise including only a residual background level and noise sources or to estimate a constant background, all sources can be selected as noise sources. However, to enable noise mapping at all, STI and/or RASTI always has to be selected since the primary use is with speech intelligibility (if all sources are selected to be noise sources phase two is not run so STI/RASTI takes no effect). RASTI (Plot-files)


Mapping of RASTI with and without background noise (from General settings Env.). The plot-file created, RASTI.PLT, shows 2D and 3D maps with and without background noise taken into account. Requires both 500 and 2k octave-bands or sum to be selected. If a 3D box is checked a separate 3D map file is created with the name RASTI_with_noise_oct_n_3D.PLT and RASTI_with_noise_oct_n_3D.PLT where also vertically overlapping audience planes can be seen. STI (Plot-files)


Mapping of STI with and without background noise (from General settings Env.). The plot-file created, STI_name.PLT, shows 2D and 3D maps with and without background noise taken into account. STI type can be selected in Preferences, depending on type chosen the name part of the file changes. If a 3D box is checked a separate 3D map file is created with the name STI_with_noise_oct_n_3D.PLT and STI_with_noise_oct_n_3D.PLT where also vertically overlapping audience planes can be seen.

57


Both direct and direct+reflected: Audience planes (Settings)

Audience area mapping

Which audience planes to map. If vertically overlapping audience planes are selected, only the highest located plane is used for 2D maps and fixed 3D maps but 3D rotatable maps show result on all audience surfaces. The selected audience planes will be marked grey in the VIEW4.PLT plot-file. To make the map statistics more useful, map points for which the associated receiver sphere significantly penetrates a wall, pillar or other object, are removed. The actual limit used is when a wall or an object is closer than half the map step. Note: if planes that do not exist in the model are selected, it will not be trapped until the mapping actually is run. Map step (Settings)


The map is divided into a rectangular grid where the square grid step can be selected. Recommended step is 0.5 - 1 m. If some part of a grid point, and thereby a fraction of the receiver sphere, should come outside the room (very close to a wall e.g.) the received energy will be too low. Direct sound mapping can have a finer resolution and the receiver is a point in the middle of each map square. Map points for which the associated receiver sphere significantly penetrates a wall, pillar or other object are removed. The actual limit used is when a wall or object is closer than half the map step. Map height above audience planes (Settings)


The map points (center of the receiver spheres) are placed at the selected height above the selected Audience planes. The height should be half a Map step above the audience planes for normal audience area mapping but for special cases they can be placed higher (e.g. factory noise mapping or a train station where the floor can be set as the "audience" plane and the height to e.g. 1.8 m). To prevent receiver spheres to go down through the audience planes the lowest allowed height is checked to be ≥ half a map step. Add direct sound with phase (Settings)


With this option selected direct sounds from each of the sound sources are allowed to interfere (two equal in-phase source contributions give close to +6dB while out-of-phase signals can be -10 dB down or more over an octave-band). Without this option the direct sound is, just like higher order reflections, added on a power basis (two equal source contributions always give +3dB). The interference calculation is based on digital octave-band filters and not only on the octave-band center frequencies (i.e. the interference is calculated with full bandwidth but the result is filtered in octave-bands so interference patterns will be somewhat smoothed but correct for the 1/1-octave band as a whole). Note: the addition with phase affects LFC values too since a figure-of-8 microphone is simulated for the lateral sound taking into account the reversed phase of reflections registered at the left side of the microphone (this is of little importance since LFC is not defined for use with multiple sources). (Octave-bands) (Settings)


The mapping is performed only for the octave-bands selected unless sum or A-weighted is included. For RASTI mapping both 500 and 2k or sum/ A-weighted have to be selected. For STI mapping all bands included have to be selected (see Preferences). To get a plot for the sum or the A-weighted sum all bands will have to be calculated even if no plots are requested for them so that it takes longer to calculate e.g. 1k and sum (calculates for 125, 250, 500, 1k, 2k, 4k, 8k, 16k unless a Preferences option excludes 8k and 16k from the sum/ A-weighted) than e.g. 1k and 2k (calculates only 1k and 2k). Also note that the Lp1m values affect D-50, C80, Ts, etc. for sum/A-weighted while for single-source single-octave maps it has no effect other than for SPL (this is not a limitation it is just how the measures work – they are early/late sound ratios). For A-weighted results they affect all measures for the sum so that e.g. D-50

58


will be the A-weighted D-50 and thus having the highest weight for 1k and 2k. A-weighted result files will wend with _asum .PLT. (Plot-files)


Since mapping does not generate any text result-files except mapping statistics, at least one of these plot options has to be selected. Contours (Plot-files)


Which directivity contours to include in the SPLcont plot. For details on directivity contours, see Section 4.3. Include source id:s (Plot-files)


Option to include source ids in map plots (useful to switch off if many sources are used not to clutter the view). PARn_oct.PLT (Plot-files) (direct+reflected)


PARDIRn_oct.PLT (Plot-files) (direct)


For each one of these optional plot-files the information as well as its location on the plot can be selected (upper left/right or lower left/right). Some of these options are available only in direct (d) and some only in direct + reflected (d+r): None

no plot created for the position;

,2D

a 2D map is created;

,3D

a 3D map is created using a static parallel projection (same as for VIEW4.PLT). In addition a separate 3D map file is created with the name e.g. D-50_oct_3D.PLT. In this 3D map also vertically overlapping mapping surfaces can be seen;

SPLdir

direct sound. Note that with more than one source the relative time delays are not seen in this plot.

SPLcont

direct sound -3, -6 and or -9 dB contours are plotted on the audience planes (see Contours group). Note that with more than one source the relative time delays are not seen in this plot. For details on directivity contours, see Section 4.3.

Closest

(d) a color-coded map of the source group with the shortest delay (and that has direct sound) electrical delays taken into account. To use this function with separate sources use source numbers A0, B0, C0 instead of A0, A1, A2 etc. Can be used together with Delay to optimize source delays. The old method with source group letters instead of colors (< v8.0e) is available via a hidden option.

Delay

(d) means that the shortest source delay is displayed (electrical delays taken into account) and is only meaningful if more than one source is used. Can be used together with Closest to optimize source delays.

RT’

(d+r) means a reverberation-time measure that relates well to EDT and is derived from Ts assuming an exponential decay (RT' = 13.8 Ts), see Theory.

(Mapping time intervals for SPL and LF) (Settings)(Plot-files)


Early sound coverage mapping can be selected in four different, possibly overlapping, timewindows. In the created plot-file view, the results for each time-window is placed corresponding to the time-limits in the dialog. Note: times are relative to arrival of the direct sound.

59

2. PREDICTION MODULE SPL (Mapping time intervals for SPL and LF) (Settings)(Plot-files)


Mapping of sound pressure level (creates SPL_oct.PLT and SPLDIR_oct.PLT ) in each selected time interval. The Scales settings for SPL affects the display scale. Tip: to map the sound that arrives first set one of the time interval to 0 - 0.1 ms. If a 3D box is checked a separate 3D map file is created with the name SPL_oct_n_3D.PLT (n is 1..4) where also vertically overlapping audience planes can be seen. LF (Mapping time intervals for SPL and LF) (Settings)(Plot-files)


Mapping of LF (creates LF_oct.PLT and LFDIR_oct.PLT ) in each time interval (i.e. the fraction of lateral to total incidence). For the lateral fraction the head-direction used is according to head-direction chosen in General settings. If a 3D box is checked a separate 3D map file is created with the name LF_oct_n_3D.PLT and LFDIR_oct_n_3D.PLT (n is 1..4) where also vertically overlapping audience planes can be seen. Scales... (Plot-files)


Auto spans the color palette as much as possible but fixed settings may sometimes be preferred so that comparisons between two calculation cases can be made. If the actual max/min values encountered are higher/lower than a selected fixed range, the max/min markers are drawn slightly above/below the color scale to indicate saturation. Note: the scale and the color palette can be changed in the View module.

Early part detailed ISM The Early part detailed ISM dialog controls image source and early diffuse reflection settings and the type of output data to be generated for detailed qualitative reflection analysis of the early part of the echogram. Direct sound and specular reflections are colored blue, first order diffuse reflections are colored red.

60


Specular refl. order (Settings)

Early part detailed ISM

Max specular reflection order (0..9). Selecting 0 will give direct sound only. Diffuse refl. order (Settings)


Max diffuse reflection order (0..1) but not higher than the specular order. I.S. truncation time (Settings)


Max impulse arrival-time for the receiver located the shortest distance from the main source. Echogram (Plot-files)


(I_ss_rr_oct.PLT). The type of echogram can also be selected. (Plot-type) Pa-scale

Early part detailed ISM/Echogram

displays the octave-band echogram in Pa simply as the square root of the rms pressure squared (which is the primary quantity calculated) times 1.414. This plot will always scale the values to set the first reflection at the Pa-axis maximum (otherwise the weaker reflections would be invisible).

Sone-scale + ear-integrated

the Sone scale plot attempts to emulate how the ear interprets the echogram, see Section 2.5. The unit will be Sone (roughly). Time-const

Early p

Time-constant for ear integration/smoothing. Default values can be selected in Preferences. Note: for "sum" echograms the 1k Hz value is used.

61


dB-scale

displays the echogram in dB where backward and forward integrated curves are optional. Note: since the late part is not calculated the echogram is incomplete and only the first part of the backward integrated decay is useful.

Show first-order surface ids

Early part detailed ISM/Echogram

Displays the plane/surface ids at the top of the bars representing first-order reflections. Refl. history (Plot-files)


Interactive display of reflection traces, echogram, source exit and receiver incidence angles (H_ss_rr_oct.PLT). View angles for the initial parallel 3D-projection used in the reflection trace view can be selected (can be changed in the View module). Note: a Preferences option can be selected to instead create a 2DA file so that complex halls can be zoomed (but not st rotated). The rays shown are colored according to reflection order: direct = black, 1 = red, nd rd th th th th 2 = blue, 3 = green, 4 = yellow, 5 = pink, 6 = turquoise, 7 = light blue. I.S. space (Plot-files)


Shows calculated image sources (S_ss_rr_oct.PLT). A reflection growth graph is shown comparing the growth in the early compared to the late part using classical theory. The reflection growth plot is only meaningful with closed rooms. (Octave-bands)(Plot-files)


Which octave-bands to generate plot-files for. Post-processing (E_ss_rr.ECH) (Save (special-case) data for)


Creates files for use with the Post-processing module, see Section 7. Filenames E_ss_rr.ECH. This saves only early reflections and should be used only for study of echoes in e.g. a stadium that is open and contains no reverberation such as in a room. To use Full detailed calculation may require too many rays to be reasonable to catch echoes coming from a distance of 150 m. On the other hand, image source echoes are likely to be exaggerated for reflection orders > 1 since no scattering is applied and should be taken only as an indication of an echo risk and the reflection path and surface sizes must be carefully examined to see if a strong echo actually would be produced in reality. Source addition (E_ss_rr.ADD) (Save (special-case) data for)


Creates files for use with the Source addition module as described in Section 6. Filenames E_ss_rr.ADD for echogram data and name.CMB for source-combination data (where name.LOC is the current source-file name). Only early reflections are saved, see note above on Post-processing.

62


Full detailed calculation

The Full detailed calculation dialog controls the settings for the randomized tail-corrected cone-tracing (RTC and an optional variant with ray-tracing for the late part). For a method description, see 2.5 Theory. Direct sound, first order specular and diffuse reflections and second order specular reflections are handled separately to be correct also for low ray/conenumbers. A separate ray/cone-tracing is performed for each octave-band (due to frequencydependent diffuse reflection) and all bands are always calculated. In addition to the numerical measures that can be selected for display below (D-50 etc.), the Stage support parameters (STearly, STlate, STtotal) [Gade-92] are calculated for any receivers that are placed at 1 m (±1 cm) and at 1 m (± 1 cm) height above the closest surface below the source. The ST results are shown only in output text-files as the last item. Number of rays/octave (Settings)


At least 5000 rays must be used for good results. With low ray-numbers the results will be too random to be useful although 1000 rays most often gives reasonable results for quick tests. See also Auto number. The more rays that are used the longer the early part of the echogram becomes (the part post-processed as is for auralization), see Theory section. While calculating, the estimated remaining time for all selected sources is shown. Auto number (Settings)


Automatic selection of a good/sufficient number of rays. This number is dependent on the

63

2. PREDICTION MODULE 2

room size and some other parameters: 1) One number is calculated to give one direct ray / m at a time corresponding to 80 ms arrival relative the direct sound for all positions. This give a good sampling of parameters such as D-50, C-80 and LF. 2) One number is calculated 2 corresponding to one ray / 4 m at the longest hall dimension. The highest of 1) and 2) is selected. 2) will only be higher for rooms that are very big so that there is some chance to get back some detailed reflections from far-away surfaces. Ray truncation time (Settings)


Truncation time in ms for rays to be traced. It is recommended to let rays continue down to -60 dB (i.e. to set ray truncation time approximately equal to the reverberation time) so that the T-30 measure can be well estimated. See also Auto time. Auto time (Settings)


Automatic selection of sufficiently long truncation time based on the estimated classical RT (only possible if rooms are closed and if classical RT does not differ too much from the one estimated by the RTC e.g. because of non-exponential decays). Late part ray-trace (instead of RTC) for special cases (Settings)


Uses an algorithm option better suited for coupled rooms and other unusual cases, see Theory. The [?] button gives a brief explanation about the cases that can benefit from using this method. Include parameters normalized to expected values (Settings)


Display some room acoustic parameters normalized with expected values according to a theory developed by M. Barron [Barron-88] in addition to absolute values, see Theory. Create one text-file/receiver (Settings)


Generation of one text-file per receiver position (E_ss_rr.TXT). Animate rays (Settings)


Makes it possible to see the actual rays traced and the emerging echogram: ·

with animation on, the calculation time increases considerably but when switched off (No) nearly the normal speed is resumed. If no animation at all is selected the speed difference from previous versions is negligible.

·

with Sequence module batch predictions, no animation is shown independent of selections.

·

the 3D model can be rotated or zoomed at any time (also while animating) using the same mouse commands as in the PLT-viewer and the 3D-viewer (LMB rotates, Ctrl-LMB or LMB+RMB shifts, RMB zooms in/out). The rotation is a bit more responsive when Step is selected but 3D manipulation can be done while calculating.

·

the 3D model always uses a parallel perspective.

·

the 3D view of the hall assumes that the coordinate system used has its origin inside or

64


close to the actual hall geometry or it can be hard to find the model (use Ctrl-LMB). ·

when rays are stepped one at a time, the current pictures can be copied and pasted into e.g. Word. Use (Copy) 3D for the 3D view of the room and (Copy) 2D for the current echogram or map.

·

for each source the rays are color-coded Red-Yellow-Green-Magenta-Blue-Black-Light grey where each color transition covers 10 dB (i.e. 0-10-20-30-40-50-60 dB, below 60 db the ray is truncated in the animation). Pure red corresponds to the strongest direct sound for each source at the receiver

·

direct and 1 order rays are drawn with heavy lines to make it easier to follow the rays.

·

if an animation is stopped (No) the calculation continues and if the animation is resumed (Yes or Step) the echogram is updated to include all previous hits.

·

only one selected receiver's echogram at one selected octave-band is animated (but all bands are calculated as usual).

·

the emerging echogram curve is drawn in blue and starts with the sum of the direct sound, all 1st and 2nd order specular reflections, and 1st order diffuse reflections (calculated without rays). The remaining reflections created by each ray are then added (drawn as red bars) while showing the continuously growing full echogram.

st

Echograms (For each receiver) (Plot-files)


Creates plot-files (E_ss_rr_oct.PLT) with calculated echograms, all estimated major parameters and a variety of other data, see section 2.4. For what is considered to be the "early part", see Theory. Note: Even if this option is off, the data can always be read in PARAM_ss.TXT as an overview for all used positions. Echogram smoothing filter (For each receiver)(Plot-files)


Filter type and octave-band dependent time-constant for the early part smoothed responses. 2 The filter acts on the p response and a small symbolic filter-function picture is shown on the E_ss_rr_oct.PLT file. Default time-constant values can be selected in Preferences. Note: for "sum" echograms the 1k Hz value is used. The filter types are: None

No smoothing filter.

Exponent.

The early echogram is smoothed by an exponential filter: with the selected timeconstant.

Rectang.

The early echogram is smoothed by a rectangular filter with width same as the selected time-constant.

Triang.

The early echogram is smoothed by a triangular filter with width same as the selected time-constant.

(Echogram/Sound Rose/Vectorgram bands)(For each…)(Plot-files) Full detailed calculation

Selection of which octave-bands to create echogram plot-files for (E_ss_rr_oct.PLT), sound roses (R_ss_rr_oct.PLT) and vectorgrams (V_ss_rr_oct.PLT). Note: even if e.g. only 1 kHz is selected all bands are always calculated.

65

2. PREDICTION MODULE Decays (For each receiver)(Plot-files)


Creates plot-files (E_ss_rr_INT.PLT) with backward integrated decays for all eight octavebands. Sound Roses (For each receiver)(Plot-files)


A plot option similar to that offered by the Early part detailed ISM history plot (plots like these are sometimes called "hedgehogs"). With this option selected six different (possibly overlapping) time-ranges can be chosen.

For each of these ranges a display of incidence angles will be created in a plot-file with the name syntax R_ss_rr_oct.PLT (R for Rose). This type of plot can be seen as a complement to the directional early echogram graphs since it also can indicate the incidence at later times. Vectorgram (For each receiver)(Plot-files)


An alternative way to display an echogram including reflection incidence direction, name syntax V_ss_rr_oct.PLT. Just like a normal echogram there is a time axis and the direct sound is placed at time zero. Reflection strength in dB is the length of each colored line. Reflection incidence direction is coded as the angle of each line (the head direction is along the time axis, if looking towards the source the direct sound is flat forward along the axis). st Color change at selected time intervals to give a depth cue. Direct sound and 1 order specular reflections are drawn with thick lines. By viewing the vectorgram from different directions different aspects can be studied. For a background to vectorgrams see [Pickover].

Note: the selected Max plot-time can not be longer than the early part. EKgrad (For each receiver)(Plot-files)


Option for an echo-disturbance criterion according to [Dietsch&Kraak-86], name syntax E_ss_rr_EKgrad.PLT. The created graphs show estimated 10% and 50% echodisturbance for speech (red) and music (blue) overlaid on the echogram for each octave (grey), for details see, see Theory. Optionally, EKgrad for speech only, music only or speech and music can be calculated.

66


(Global statistics)(Plot-files)


Various statistics based on the results for all used receivers. Absorption (Global statistics)(Plot-files)


Absorption coefficient histograms, ABS_ss.PLT (125Hz to 16k Hz on one plot). Free Paths (Global statistics)(Plot-files)


Free path histograms, PATH_ss.PLT (125Hz to 16kHz on one plot). Wall hits (Global statistics)(Plot-files )


Wall hit statistics, HITS_ss.PLT. RT overview (Global statistics)(Plot-files)


Graphs of estimated reverberation times, mean absorption coefficients, rays lost/absorbed and other statistical data (RT_ss.PLT). An option to include a reference or target RT curve is given. Note: the Ref RT can be given with two decimals but that does not mean that prediction accuracy is two decimals. (Overview)(Plot-files )


Various plot-files showing results for all receivers either as graphs or numbers placed over a hall plan. Overview plot-files are created only if more than one receiver is used. Parameter map (Over all receivers (if > 1))(Plot-files )


Acoustic parameter overview plot-file, PARAMn_ss.PLT (n = 1,2,3). Which parameters to plot for all receiver positions can be selected. The plot will be with four plans of the hall and four or less parameters on one page. The abbreviations are the same as are used in the text resultfiles, see Section 2.4.

67

2. PREDICTION MODULE Parameter trend (Over all receivers (if > 1))(Plot-files )


Acoustic parameter overview plot-file, TRENDn_ss.PLT (n = 1,2,3). Parameters: see above.

Normalized trend (Over all receivers (if > 1))(Plot-files )


Similar to trend but with normalized values (if Include parameters normalized to expected values is on). The following parameters are shown: D-50, C-80, G, SPL. Filename NTREND_ss.PLT. For a background to normalized parameters see Theory. RASTI (Over all receivers (if > 1))(Plot-files)


Graphs of RASTI values, 500Hz/2kHz average D-50 (Deutlichkeit), and 500Hz/2kHz SPL (RASTI is built from these two octaves). Filename RASTI_ss.PLT. Note: RASTI is always calculated (even if this plot option is not selected or no echogram plots for 500 Hz and 2 kHz are requested) and can always be found in PARAM_ss.PLT. STI (Over all receivers (if > 1))(Plot-files)


Graphs of STI and octave band TI (Transmission Index) with and without background noise taken into account. The STI type or octave band weight values used can be selected in Preferences, see Section 1.3. Filename STI_ss.PLT. Note: STI is always calculated (even if this plot option is not selected or no echogram plots for all bands used in STI are requested) and can always be found in PARAM_ss.PLT. SPL spectra (Plot-files)


An optional plot SPECTRA_ss.PLT plots the SPL octave-band spectra for direct sound, first 50 ms and total as an aid in overall system eq and coverage. Spectra for all receivers are shown in light grey and each receiver’s data is as plot frame (in a 2DA plot-file) also indicating a line from the source to the receiver (dashed if there is no direct sound). A reference or target spectrum can be given for reference (plotted as rings with a 1 dB radius). Post-processing (E_ss_rr.ECH) (Save data for)


Creates files for use with the Post-processing module to create auralization, see Section 7. Filenames E_ss_rr.ECH. In the full software version, a file called Walker.CAG is created that can later be used by CATT-Walker. Source addition (E_ss_rr.ADD) (Save data for)


Creates files for use with the Source addition module as described in Section 6. Filenames E_ss_rr.ADD for echogram data and name.CMB for source-combination data (where name.LOC is the current source-file).

68


Interactive RT estimate For this function the geometry is read according to the current prediction General settings and a dialog and a dialog where the Sabine and Eyring RT values can interactively be evaluated based on surface property changes, in addition a quick global ray-tracing can be performed.

By changing a property assigned to a plane or sub-plane the classical RTs are immediately recalculated along with the mean absorption coefficient. If the Surface properties module is open double-clicking in the property list automatically changes the selected plane's property in the Interactive RT estimate dialog and updates the RT estimates. By clicking the button with the current property name (SOMEWOOD above) one of the GEO-defined ABS properties (as well as the predefined ones) can be selected from a list. Planes that have been altered have their names colored differently and the change can be Undone. After OK is pressed a text-file is optionally created listing the changes and the resulting RTs. This file can be used when editing the GEO-files to implement the changes selected. Note: the Eyring RT estimated here is not exactly corresponding to the EyrT and EyrTg measures shown after Full detailed calculation. EyrT and EyrTg both use the mean free path estimated by the rays rather than 4V/S. EyrT uses the mean absorption coefficient estimated by the rays while EyrTg uses the classical area-weighted coefficient.

69

2. PREDICTION MODULE Sort order (Plane list)

Interactive RT estimate

The list of planes in the model can be sorted in ascending order according to: Plane id Plane name Area Key name Abs. 125&250 Abs. 500&1k Abs. 2k&4k Abs. 125-4k Abs. area 125&250 Abs. area 500&1k Abs. area 2k&4k Abs. area 125-4k

plane ids (sub-sorted on sub-division names) plane name (sub-sorted on sub-division names) plane/sub-division area absorption property name absorption coefficients in the 125 and 250 Hz bands 500 and 1k bands 2k and 4k bands 125 to 4k bands (overall absorption) absorption area in the 125 and 250 Hz bands (m²Sabine) 500 and 1k bands 2k and 4k bands 125 to 4k bands (overall absorption area)

Save statistics


Creates a file listing all used absorption properties sorted after the most used material together with their total areas and relative absorption area (the Used column lists how many surfaces uses the property). Settings = C:\CATTDOCS\v8\Example\Catt.PRD PROJECT = v8.0 Manual Example Hall MASTER = MASTER.GEO The following planes are changed: 20 stage wall from SOMEWOOD to BRICK50 The absorption is used as follows: ----------------------------------------------------------------------------------Relative absorption area [%] Key name Area [m²] Area [%] < 125 250 500 1k 2k 4k 8k 16k> Used ----------------------------------------------------------------------------------SOMEWOOD 1859.0 61.6 < 54.7 39.3 28.1 22.6 16.6 12.5 7.1 2.3> 13 AUD 596.1 19.7 < 30.7 44.5 54.7 58.0 60.2 60.0 60.5 65.3> 5 ABSORBER 272.0 9.0 < 8.0 11.5 13.7 16.5 20.9 25.6 31.0 31.5> 5 DIFFUSOR 216.0 7.2 < 6.4 4.6 3.3 2.6 1.9 1.4 0.8 0.3> 1 BRICK50 75.9 2.5 < 0.2 0.2 0.3 0.3 0.4 0.5 0.6 0.6> 1 ----------------------------------------------------------------------------------3019.1

Start trace (RT estimate)


Starts a global ray-tracing that continuously updates a global T-30 estimate. The benefit is that the effect of diffuse reflection (or rather lack there of) can be seen already before using Full detailed calculation. If the room is non-mixing with uneven absorption distribution, T-30 will be longer than the Sabine and Eyring values. Since this is a global ray-trace the resulting T-30 for diffusion-sensitive cases is slightly lower than that estimated at single receiver positions by Full detailed but gives in principle the same result as an average over many receivers. The ray-tracing is always performed using an omni source and rays are continuously sent out uniformly from one of the sources defined (which one does not really matter for a global raytrace unless the room shape is very special and a global ray-trace is of less value anyway). Every 100 rays per octave T-30 is estimated and updated and gradually goes toward a stable curve, if the curve does not stabilize it indicates that that the model has estimation problems. After 3000 rays per octave the ray-tracing is automatically stopped (a global ray-tracing can get by with fewer rays than when receivers have to be hit). The ray-tracing can also be stopped before that point by pressing STOP tracing or by changing an absorption value (the

70


ray-tracing then has to be restarted). The T-30 results should not be over-interpreted if the room has unusual features such as coupled volumes and is mainly to be seen as a way to faster find good settings for a longer Full detailed calculation and a more realistic "what if" RT estimates than what only Sabine and Eyring can offer. Copy graphs (RT estimate)


Copies the RT graphs to the clipboard for pasting into a document. Sabine (RT estimate)


The estimated Sabine RT including air absorption as selected in General settings. See note below. Eyring (RT estimate)


The estimated Eyring RT including air absorption as selected in General settings. See note below. Undo


Undo the changes on the current plane (sets the initial property). Undo all


Undo all changes and revert to initial properties.


The estimated classical area-weighted mean absorption coefficient (AbsCg). For small values of the Eyring RT approaches the Sabine RT.


The estimated classical mean free path 4·V/S (V is estimated volume and S is estimated surface area). Volume


The estimated hall volume (V). In some cases when the volume cannot be estimated, a manual entry can be made and this text box is changed to an edit-box, press the revealed Update button after volume entry. Schröder freq.


To aid in estimating from which band geometrical acoustics (GA) may be valid, the so called Schröder frequency [Kuttruff-80] is calculated: fs = 2000

TV

where T is taken as the average of the RT at 125 and 250 Hz (initially the Sabine RT but if tracing is made T-30 is used instead) and V is room volume. However, fs itself gives a much too low estimate of the GA limit so an empirical 4fs is used instead, see below. -> Geom. acoustic estimated from band Interactive RT estimate


An estimate of from which octave-band geometrical acoustics (GA) may be valid based on an

71


empirical 4fs where fs is the Schröder frequency, see above. The resulting band matches well the general experience from modeling small and large rooms but can still only be seen as a rough guideline. The frequency limit of GA is not only dependent on the modal density and overlap as fs indicates but as well on surface sizes. If the results are saved (OK button) fs and estimated GA limit are also given in the result text-file. Update


Updates the RT estimate after a manually entered volume (only if the room is not considered consistently closed) Note: Sabine or Eyring RT estimates are only valid in rooms that either have "mixing" shapes, very even absorption distribution or that have significant diffusion. With a Full detailed calculation the validity of the Sabine and Eyring estimate can be seen (if the T-30 values are close to the Sabine values, a considerable amount of time can be saved by first using the Interactive RT estimate calculation). Note also that the geometry must be correctly modeled to give accurate enough volume and surface area estimates and it must not be judged to be an open model, see Section 2.1

Pixel rendering For this function the geometry is read according to the current prediction General settings and a dialog with a room geometry 3D wire-frame is shown.

The direct sound rendering is in principle the same as the Direct sound only Audience area mapping but mapping is made directly on all surfaces and since only the direct sound is traced, and no vector graphics results are created, it can be implemented differently. For example, it would be impossible to map D-50 or C-80 in a similar way since then an infinite number or rays would be required. Semi-transparency is taken into account in this view as long as the sound does not have to pass through more than one semi-transparent surface. To rotate, zoom and shift the model the same mouse commands as for 3D PLT-files and the 3D-viewer are used, the dialog can also be resized (grab a dialog edge, the initial size is the smallest allowed). As soon as a new 3D view, window size or source combination is selected

72


a rendered bitmap is replaced by a wire-frame. The idea is that interesting views are to be copied to say a Word document using the Copy button. The reason for this is that since this is pixel rendering, and not vector graphics like the 3D PLT-files, the resolution will anyway not be higher than shown in the view if it were to be printed on a high-res printer, this is similar to the OpenGL 3D-viewer that also uses is a pixel rendering technique. Model (Render)

Pixel rendering

Renders all model surfaces. Aud. (Render)

Pixel rendering

For cases when rendering of all surfaces is not interesting, mapping on only the fictive audience surfaces can be made and can be considerably faster (audience planes and mapping height are taken from Audience area mapping with direct sound only). Fictive audience surfaces are outlined with dotted black lines to indicate that they are not an actual part of the model geometry. Note: since a fictive audience plane has no absorption, the Refl. sound button is disabled if the Aud. button is enabled. Geometry (Render)

Pixel rendering

The current view is rendered on a pixel by pixel basis using a slight incidence dependent color shading as if a light came from the viewing direction. This is a simple rendering that is fast for most geometries and its main purpose is to check the model for errors. This rendering uses the same equations to solve ray/plane intersections as are used in ray- and cone-tracing so if planes are warped and similar an error in rendering should show up also here (the Colored.PLT plot-file uses a flat shading technique that does not use rays so errors do not show up in the same way). Calculation time: this rendering uses a single main stage and the time depends on the complexity of the room and how many pixels that need to be rendered (window size and how much of that window the model occupies) Direct (Sound)(Render)

Pixel rendering

The current view is rendered where each visible wall pixel is colored according to the SPL of the sum of the direct sound of all sources that can reach that point on the wall. A subset of sources defined in General settings can be selected by the Sources button. Points that no source can reach are colored black indicating direct sound shadow-zones. Very weak sounds that are at or below the lowest scale value are colored dark blue and can thus be distinguished from no direct sound at all. Calculation time with power addition: this rendering uses two main stages, "rendering" and “source:addition". The time depends on the number of nonempty pixels, the room complexity, the number of sources and source types (SD2 arrays generally takes more time). Calculation time with interference: this rendering uses three main stages. The first two are near identical to power addition but a third stage is required calculating the interference, this last stage depends on the number of pixels visible to at least one source, and number of sources only. Reflected (Sound)(Render)

Pixel rendering

this is identical to Direct but instead of the sound that hits a wall point, the sound that is reflected off that wall point is displayed, i.e. the incident sum of sound multiplied with (1-α). If the surface is semi-transparent the factor is instead (1 - α)(1 - τ). Calculation time: same as for direct sound. (Max, dB)(Sound)(Render)

Pixel rendering

The scale Max can be chosen (will be rounded upward in 5 dB steps), if the max value at any visible pixel is higher, a black line is drawn slightly above the scale, if the actual max encountered is lower than the scale max it is indicated by a white line across the scale, the

73


lowest value is indicated in the same way at the lower end. If the ? button is clicked a quick estimate of the max SPL in the current view is performed. The resulting value cannot be guaranteed to be the absolute max since it is calculated from a coarse map but if not, the actual max from the latest action is indicated numerically (to the right of the chose Max) so for the next run a better max can be selected. Cases where the estimated max may not work is e.g. when a not very directive source sits very close to a wall and a wall pixel is visible, the SPL there is then high but may not be found by a coarser map (but it is also not very interesting to use that as max). The max estimate goes ca. 25 times faster than a full render. As long as CATT-A is not restarted the max remains the same also when closing/opening the dialog. Range (Sound)(Render)

Pixel rendering

The scale Range can be chosen i.e. from Max and down. For sound systems 30 dB is a good range while for omni and less directive sources 10 or 20 dB works better. As long as CATT-A is not restarted the range remains the same also when closing/opening the dialog. Band (Sound)(Render)

Pixel rendering

The octave-band for the rendering, as long as CATT-A is not restarted the band remain the same also when closing/opening the dialog. Interference (Sound)(Render)

Pixel rendering

If Interference is used coh (coherent) will be indicated below the color scale otherwise pwr. Interference has no effect with one source only and pwr will be indicated. When Interference is selected the source contributions are added with phase and just like in Audience area mapping (Add direct sound with phase) the interference over the complete octave-band is calculated (i.e. not just for the center frequency) using octave-band FIR filters (and hence takes some time). Recall/Store (3D-views)

Pixel rendering

Three options for storing/recalling 3D views (view angles etc.). As long as CATT-A is not restarted these remain also when closing/opening the dialog. Sources

Pixel rendering

Option for selecting a subset of the sources selected in General settings. If some sources have been switched off it can also, via an option when the dialog is closed, affect the selection in General settings. Copy

Pixel rendering

Copies the current bitmap into the Windows clipboard for pasting into e.g. MS Word or Paint.

Pixel rendering application tips Since some if these rendering operations can take several minutes to perform if there are many sources, the following procedure is recommended for most efficient use: ·

first select a good wire-frame 3D-view of the surfaces that are interesting, leave some space at the right for the color scale. Note that the dialog can be resized but a larger map will also take more time. Possibly store some good views for later recall when comparing after a geometry or source change. The last selected 3D view, dialog size, scale and octave-band will be selected the next time the dialog opens to enable comparisons using an identical view. Also all stored views will remain the same and can be recalled as long as

74

2. PREDICTION MODULE CATT-A is not restarted.

·

optionally perform a quick Geometry rendering

·

perform (Sound) renderings without Interference to set the color scale and perhaps adjust the view. To get a good initial start for the max scale value press the ? button.

·

If desired, switch on Interference and render the final view.

·

Copy the view to a another application such as Word or Paint, go back to the first step and select another view.

·

Select Close when you are finished or want to change room or source details. If the source combination has changed, General settings can optionally be updated accordingly.

75


2.3 Input-files The prediction input-files describe the geometrical model, source positions/properties, and receiver positions.

Common features All geometry-related input-files are in text-format: (*.GEO, *.LOC) and the data entry is done using any suitable editor or via an export from the AutoCAD interface, DXF-conversion via DXF2GEO or via a modeling plugin, see chapter 10. CATT-Edit is recommended since the main program then knows if a file has been changed before a prediction is made and it also has several other useful features such as syntax templates. In the data-file syntax descriptions below the following apply: UPPERCASE COURIER BOLD is used for directives

lowercase courier is used for values to directives UPPERCASE COURIER is used for filenames and similar items |

means "OR"

...

means a sequence, range, or continuation of the specified item(s).

Text like this with a left border is a syntax description. The following applies to all text input data-files: ·

character case is not significant except for plane and sub-division names in geometry-files and TEXT marker text which will stay as entered

·

data is entered and parsed on a line-by-line basis. An exception to this is definition of planes in geometry-files that can be spread over many lines

·

totally blank lines are ignored allowing for “white space“ in the files

·

data items must be separated by one or more blanks or TABs

·

no blank or TAB is, however, needed between a data field delimiter and a data item

·

anything written after a semicolon “;“ on a line is treated as a comment

·

reserved characters (that cannot be used in constant and absorption names etc.): * ; [(< >)]/\- +%

·

entering of floating point values has a very free format. The floating point number 1.0 can, for example, be entered as any of the following: 1 1. 1.0 1E+00 1.E+00 1.0E+00 1E0 1.E0 1.0E0

·

both decimal point and decimal comma are allowed.

76


·

negative numbers (or constants) must be entered without a space between the minus sign "-" and the first digit (i.e. -1.0 is OK but - 1.0 is not)

·

unit for coordinates etc. is meter [m]. Note: see the SCALE directive for using other units

·

names (constant, source directivity names and absorption/diffusion properties) must not have more than 15 characters, consist only of letters, numbers (0..9), and underscores ( _ ) and the name must start with a letter

·

if a directive is left out the default is used

·

if a directive is without meaning in a particular context it is ignored

·

should by mistake two directives of the same type (that are not defined to be multiple directives) be entered, it will be trapped. If a directive is defined as a multiple directive, every appearance counts

·

IF-THEN-ELSE statements (ELSE optional): IF expr1 condition expr2 THEN statements ELSE statements ENDIF

or IF expr THEN statements ELSE statements ENDIF

where expr1, expr2, and expr are valid expressions. The condition can be any of: <, <=, = , >=, >, and # corresponding to less, less or equal, equal, greater or equal, greater, and not equal. If expr evaluates to > 0 then the statements are executed. To enable high-level statements two global constants have been pre-defined:

FALSE = -1 and TRUE = 1. Example: if a and b are declared as either TRUE or FALSE then: "IF a THEN"

can be written as

IF a THEN

"IF NOT a THEN"

can be written as

IF -a THEN

"IF a AND b THEN"

can be written as

IF a+b THEN

"IF a OR b THEN"

can be written as

IF a+b+1 THEN

"IF a XOR b THEN"

can be written as

IF -a*b THEN

Note: IF-ELSE statements cannot be nested. ·

SAY statements: SAY message

77


will display message on the screen and wait for OK to be pressed. Can be used for tracing and debugging purposes. SAY constant

where constant is the name of a previously declared constant, will display the evaluated value or string and wait for OK to be pressed.

· DEBUG statements: DEBUG message|constant

Similar to SAY but if DEBUGOFF the message is just written to the Debug window and no message box is shown. DEBUGON

If either OK or Cancel is clicked the parsing continues but clicking Cancel sets debug off. In both cases the result is added to the Debug window. Debug is off when the Master file is beginning to be read. DEBUGOFF

Makes DEBUG messages be displayed only in the Debug window and the parsing continues. ·

RETURN statements: RETURN

will force end-of-file. Can be used for debugging purposes. ·

BREAK statements: BREAK message

will force abort of the processing with the message shown on screen. Example:

IF height >= 15 THEN BREAK Height must be less than 15! ENDIF

Geometry-files Geometry-files (GEO-files for short) describe the hall geometry and often also defines wall absorption/diffusion properties. GEO-files can be organized hierarchically like a "tree" to make structuring of the geometry possible since each GEO-file can INCLUDE one or more GEO-files which in turn can INCLUDE others and so on to any depth. An included file can have any name but must have the extension .GEO and be located in the input folder. Planes should not refer to a corner id in a file at a lower level (an included file). One of the advantages of the tree-structure is that a complete section of the hall (e.g. all reflectors) can be switched off just by adding a semicolon (comment introducer) in front of the corresponding INCLUDE directive or by using an IF-THEN statement.

78


The master-file name which the Prediction module looks for in the input folder typically has the name MASTER.GEO. By selecting another master-file (using the General settings/Master GEO-file) it is possible to have several different "setups" for a hall where e.g. one master-file includes reflectors and another just the plain hall thus making comparisons with/without reflectors easy. A GEO-file can be defined as an OBJECT and be rotated around any point and translated to any location in the hall. Via the COPY directive multiple copies of an object can be created at different positions and rotations. A GEO-file can use named LOCAL and GLOBAL constants, expressions, and IF-THEN statements, which make changes easy and secure. For interactive constant declarations GETLOCAL and GETGLOBAL can be used. A GEO-file can via the MIRROR directive be mirrored for automatic creation of symmetrical or nearly symmetrical parts. Several tools for accurate definition of plane corners are provided (x(), y(), z(), lock(), cut(), loop()).

Syntax In addition to the common features of input text-files, for the GEO-file syntax explanations the following apply: directives are entered as: DIRECTIVE value(s)

or DIRECTIVE

for toggle (on/off) directives;

directives may be entered in any order but a constant has to be declared before it is used; coordinates, scale, object origin, shift, rotation and absorption/scattering coefficients can be specified as a floating point number, as a declared local or global constant, or as an expression. A GEO-file must have three (optionally four) sections with the following order: directives constant and absorption declarations etc. CORNERS mandatory corner definitions optional PLANES mandatory plane definitionsoptional MARKERS optional marker definitions optional Note: a mandatory section may be empty but the strings CORNERS and PLANES must always be present.

79


Directives section CATT-Edit has a Help menu item with short syntax descriptions to serve as a quick reference for the detailed descriptions below.

(multiple) (short form)

INCLUDE filename #I filename

The files are included in the order they are encountered, but not until the current GEO-file is read. This means that each level of files are read before the next level. If a problem with file inclusion should occur insert SAY statements to trace GEO-files or use Utilities|Create GEOfile Tree where GEO-files are listed in order of inclusion.

filename:

name of GEO-file (with or without .GEO) to include (must be in the input folder and be given without path). Default: no file included.

SCALE sc_x sc_y sc_z

Scale to use in the GEO-file. If scaling is used all coordinates in the file will be in scaled units. If the same scale in all files is required, global constants can be defined in the master-file (e.g. sx, sy, sz) and then SCALE sx sy sz can be put in all included files as well as source- and receiver-files. Note: if SCALE is used in all GEO-files as well as in the receiver- and sourcefiles using x(), y(), and z() functions may scale twice since scaling has already been applied. In such a case use e.g. 1+x(12)/s, where s is the scale factor, to prevent the extra scaling.

sc_x, y, z:

scale factors for each axis. E.g. all x-values will be multiplied with sc_x. Default: SCALE 1 1 1 (i.e. no scaling)

OBJECT

To make the GEO-file be an object-file. The coordinate system in the file is then local to the object. The sub directives ORIGIN, TRANSLATE, and ROTATE are used to place the object in the hall system. Default: not an object-file. TRANSLATE t_x t_y t_z SHIFT t_x t_y t_z

(alternative form)

How much to translate (shift) the object - i.e. where, in the hall coordinate system, to put its origin. See Fig. 2.1.

t_x, t_y, t_z : translation along each axis. Default: TRANSLATE 0 0 0 ROTATE r_x r_y r_z

How much to rotate the object around its origin. See Fig 2.1.

80


r_x, r_y, r_z : rotation around each object axis in degrees. A positive value rotates clockwise and negative value rotates counter clockwise. Default: ROTATE 0 0 0

y

Y r_z # 0 with new ORIGIN

y

x r_z # 0

new ORIGIN o_y > 0

x

t_y r_z = 0

Example in 2D. X-Y is hall system x-y is object system X

t_x

Fig. 2.1 Function of ROTATE and ORIGIN OBJECT directives ORIGIN o_x o_y o_z

If OBJECT. Where the object’s origin should be placed relative the coordinates specified in the file. Rotation and scaling are performed around this origin. By using this directive it is possible to enter a symmetric object in a suitable coordinate system and then rotate it around a point away from the center. See Fig. 2.1.

o_x, o_y, o_z: position of new origin in the object coordinate system. Default: ORIGIN 0 0 0 OFFSETPL number

Offsets all plane ids in the file with number. Enables a local plane-numbering system; especially useful for objects.

number:

number to add to plane ids in the GEO-file. Can be negative. Default: OFFSETPL 0

OFFSETCO number

Offsets all corner ids in the file with number. Enables a local corner-numbering system; especially useful for objects.

number:

number to add to corner ids in the GEO-file. It applies to corners in both the corners and the planes section. Can be negative. Default: OFFSETCO 0

81


FROMFRONT|FROMBACK

If the plane corners in the PLANES section are entered in clockwise order as seen from the front (reflecting side) or from the back (non-reflecting side). This is further described below about the planes section. Default: FROMBACK MIRROR coadd pladd MIRROR coadd pladd EXCLUDE plid ...

(form 1) (form 2)

The first form of this directive will automatically create a mirror copy of all corners defined in the file with non-zero x. It will also create mirror copies of all planes that have corner coordinates with the same x sign or that are zero. The mirror operation takes place after all corners are read which means that mirrored corners cannot be referred to in the corners section. Help corners for locking can be created with x = 0.

coadd:

a number added to all corner ids for mirrored corners.

pladd:

a number added to all plane ids for mirrored planes.

EXCLUDE

option to exclude on or more plane ids that will not be mirrored.

plid ...:

a list of planes ids to exclude. The list of ids can include ranges from-to (e.g. EXCLUDE 5 10-15 18 20-26), ranges can be overlapping and in any order. These plane ids correspond to those defined in the file and any used OFFSETPL will be handled automatically.

An example: OFFSETCO 200 OFFSETPL 20 MIRROR 1000 100 CORNERS 0 ... 1 ... 2 ... ... PLANES [0 ... [1 ... [2 ... ... The part explicitly modeled in the file will have corner ids 200, 201, 202 ... and plane ids 20, 21, 22 ... The mirrored part will, however, get corner ids 1200, 1201, 1202 ... and plane ids 120, 121, 122 ... The second form allows for some planes being excluded from mirroring. An example:

OFFSETCO 200 OFFSETPL 20 MIRROR 1000 100 EXCLUDE 0 3 CORNERS 0 ... 1 ... 2 ... ...

82


PLANES [0 ... [1 ... [2 ... [3 ... ... The part explicitly modeled in the file will have corner ids 200, 201, 202 ... and plane ids 20, 21, 22, 23 ... The mirrored part will, however, still get corner ids 1200, 1201, 1202 ... and plane ids 121, 122 ... but not planes 120 and 123 since planes 0 and 3 have been excluded from mirroring. The EXCLUDE option can be used if models are not perfectly symmetrical since all corners are mirrored and where some planes will be slightly altered (e.g. a door only on one side). COPY coadd pladd o_x o_y o_z t_x t_y t_z r_x r_y r_z [EXCLUDE plid…] (multiple) #C coadd pladd o_x o_y o_z t_x t_y t_z r_x r_y r_z [EXCLUDE plid…] (short form)

This directive will create a copy of the data in the file. The procedure will simply be to model one object, and then add a number of this directive for the GEO-file to copy itself. The directive also works together with MIRROR and creates a copy also of the mirrored part. The GEO-file does not have to be an OBJECT itself for COPY to work.

coadd:

a number added to all corner ids for copied corners.

pladd:

a number added to all plane ids for copied planes.

o_x ...:

new object origin (same function as ORIGIN)

t_x ...:

location of new object (same function as TRANSLATE)

r_x ...:

rotation of new object (same function as ROTATE)

EXCLUDE

option to exclude on or more plane ids that will not be copied.

plid ...:

a list of plane ids to exclude. The list of ids can include ranges from-to (e.g. EXCLUDE 5 10-15 18 20-26), ranges can be overlapping and in any order. These plane ids correspond to those defined in the file and any used OFFSETPL will be handled automatically.

An example (not using the optional EXCLUDE):

OFFSETCO 200 OFFSETPL 20 MIRROR 1000 100 ; co pl orig. COPY 10 5 0 0 0 COPY 20 10 0 0 0 COPY 30 15 0 0 0 CORNERS 0 ... 1 ... 2 ... ... PLANES [0 ... [1 ... [2 ...

transl. 0 10 0 0 20 0 0 30 0

rotate 0 0 15 0 0 30 0 0 45

83


... The part modeled in the file will have corner ids 200, 201, 202 ... and plane ids 20, 21, 22 .... The symmetric part will get corner ids 1200, 1201, 1202 ... and plane ids 120, 121, 122 .... The first object copy will have corner ids 210, 211, 212 ... and plane ids 25, 26, 27 ... The symmetric part will, however, get corner ids 1210, 1211, 1212 ... and plane ids 125, 126, 127 ... and so on for the second and third copy. The use of COPY is sometimes not directly possible if some of the copies are to be slightly different. An alternative is then to still use the COPY command copying also unwanted planes and afterwards in General settings exclude those planes (Plane use, Not some). This solution is not very elegant since a change in OFFSETPL in the object GEO-file will necessitate a change also in the General settings (easy to forget and makes the GEO and the PRD dependent). Instead, the EXCLUDE option can be used, an example:

... OFFSETCO 1000 OFFSETPL 100 COPY 5 5 COPY 10 10

0 0 0 0 0 0

0 4 0.5 -3 6 0.5

10 5 0 10 5 0

EXCLUDE 0 ;excludes side0 EXCLUDE 1 ;excludes side1

CORNERS 0 -a -b 1 -a b 2 a b 3 a -b

0 0 0 0

PLANES [ 0 side0 / 0 1 2 3 / reflabs* ] [ 1 side1 / 3 2 1 0 / reflabs* ] (multiple) (short form)

GLOBAL name = expr|string #G name = expr|string

Definition of a global named constant - the constant is then known in the current GEO-file and in every GEO-file included in and after the current file as well as in source- and receiver-files. An attempt to re-declare a global constant will be trapped. The equals sign is optional. Two global constants are predefined: TRUE=1 and FALSE=-1.

name:

constant name (only the first 15 characters are significant).

expr:

an expression, arbitrarily complex, including declared constants, parenthesis, and calls to math functions: sin(), cos(), tan(), arctan(), exp(), log() (natural logarithm), exp10(), log10()(10-based logarithm), sqrt(), sqr(), cube(), abs(), rnd(), rnd2(). Angle arguments must be in degrees. For details about function, see Use of Expressions below.

string:

a string constant in double quotes that can be used e.g. in plane definitions and the name is then delimited with %, see below. To avoid that %, as used in a message, (e.g. in the GETGLOBAL INFO etc.) causes a search for string substitution assuming %name% the % sign can be preceded with a backslash like e.g. in "Absorption coefficient in \%".

84


Example 1:

GLOBAL a = 5 GLOBAL h_ceil = a + 1 GLOBAL wup = "Wall with unknown plaster" ... CORNERS 1 a 2.5 h_ceil ... PLANES [ 1 %wup% / 2 4 5 6 7 / wallabs ] ; will expand %wup% to Wall with unknown plaster Example 2:

GLOBAL blue = "{0 0 255}" ABS plastic = <12 10 8 7 6 5> %blue% ... Since an IF-statement in the form "IF expr THEN" will be true if expr evaluates to > 0, flags can be defined by using TRUE and FALSE :

GLOBAL balcony = TRUE ... IF balcony THEN ... ELSE ... ENDIF


LOCAL name = expr|string #L name = expr|string

Definition of a local named constant - the constant is then known in the current GEO-file only. An attempt to re-declare a local constant in the same file will be trapped. If the local name should be the same as a global name the local overrules the global. The equals sign is optional.

name, expr, string:

see GLOBAL.

GETGLOBAL name = expr|string [INFO = "info"] [MIN = min | CLAMPMIN = min] [MAX = max | CLAMPMAX = max] (multiple) GETLOCAL name = expr|string [INFO = "info"] [MIN = min | CLAMPMIN = min] [MAX = max | CLAMPMAX = max] (multiple)

Same as GLOBAL and LOCAL above but the constant value or expression is entered at runtime. expr|string is the default value and optional checks on the value can be performed. Note: the syntax above spreads over two lines but actually entered text must be on one line.

name, expr, string: INFO

see GLOBAL.

is used to tell the user information about the value to enter, e.g.:

GETGLOBAL h = 10 INFO = "Ceiling height in meter"

85


or for a string constant :

GETGLOBAL s = "name" INFO = "Plane name" MIN and/or MAX

are used to check that a constant value is within limits, e.g.:

GETGLOBAL h = 10 INFO = "Height in meter" MIN=5 MAX=18 CLAMPMIN and/or CLAMPMAX

are used to check that a constant value is within limits, and if not the value is clamped to the given min and/or max values, e.g. :

GETGLOBAL h=10 INFO="Height in meter" CLAMPMIN=5 MAX=18 If e.g. 0 is given h will be set to 5 and to notify the user the *DEBUG* window will display “LOCAL h clamped at min 5.000”.

ABS ABS ABS ABS

absname = < α[/τ]−values > [{R G B}] (all ABS variants are multiple) absname = < α[/τ]−values > m < s-values > [m_data][{R G B}] absname1 = absname2 absname1 = absname2 m < s-values > [m_data] [optional]

Declaration of a global named surface (absorption/diffusion) property with optional semitransparency as an alternative to put the coefficients directly or to use the surface properties library (the fourth variant complements a previously defined property with scattering coefficients – or redefines its current coefficients). The absorbing/diffusing property is then known from this point in the current GEO-file and in every GEO-file included in and after the current file. An attempt to re-declare a property will be trapped (it may, for example, already be declared in the surface properties library). There are two predefined absorption properties: TOTREF and TOTABS. TOTABS is an artificial total absorption used for the back side of "floating" audience planes and similar planes for quick tests but should be use very seldom (better to use 99,9%). The equals sign is optional.

absname:

name of absorption/diffusion property.

<α[/τ]-values> : absorption coefficients in % ]0..100[ for the six octave bands 125 to 4k Hz (can be specified by constants): <α125 α250 α500 α1k α2k α4k> e.g. <5 7 8 10 15 20>. 8k and 16k values are extrapolated from 2k and 4k. Optionally 8k and 16k are given preceded by a colon: <α125 α250 α500 α1k α2k α4k : α8k α16k > e.g. <5 7 8 10 15 20 : 35 50 >. Optionally τ-values for transmission coefficients in % ]0-100[ for the six (or eight) octave bands can be given (can also be specified by constants): e.g. <5/55 7/47 8/38 10/30 15/25 20/10>. For guidelines regarding which types of α- and τ-values to use with semitransparency, see Theory section 2.5

m:

diffusion model letter.

L = Lambert model (no other models implemented). :

scattering coefficients in % ]0..100[ for the six octave-bands 125 to 4k Hz

86


(can be specified by constants). e.g. <5 10 20 25 30 35>. 8k and 16k values are extrapolated from 2k and 4k. Optionally 8k and 16k are given preceded by a colon: e.g. <5 10 20 25 30 35 : 40 50>. The scattering coefficients are defined according to [ISO-scatt]. To simplify the estimation of scattering coefficients of rough surfaces, modeled as flat planes, a function estimate() can be used instead of explicit coefficients, e.g.:

ABS ceiling = <15 10 11 12 12 12> L < estimate(0.4) > where the argument given (numerical value only) is the roughness scale in meter (e.g. a corrugated ceiling with 0.4 m wells). 0.4 as in the example will estimate the following values: <19 27 38 54 76 99 : 99 99> based on roughness to wavelength ratio. If the values turn out < 10% or > 99% they are clamped to the respective limit. The estimated values are no absolute truths and are just based on experience and a reasonable frequency dependence (the actual function used is s = 50·sqrt(d/λ) where d is the given value in meter and λ the wavelength). In the PLINFO plot-file, and in the a list of planes if selected in a debug file, it is indicated if scattering coefficients were estimated using estimate() or given explicitly.

m_data:

a sequence of diffusion model dependent values. Not required for the Lambert model (previous versions have had additional Lambert data that no longer are used, future models may need more data).

{R G B}:

optional Red Green Blue values in the range [0..255] for defining the color of the material (e.g. pure Red is {255 0 0}. For non-pure colors the values can be selected in a color dialog using CATT-Edit (Template|RGB Color menu). If not given a color is auto-generated. Tip: symbolic color names can be created using string constants: LOCAL red = “{255 0 0}“ and used as %red%. Note: don't use black {0 0 0} since this color is used to identifying the back side in PLINFO.PLT frames.

Notes: transparency is intended for thin cloths and curtains (i.e. not walls) and when used: ·

the semi-transparent planes: have to be interior (such as hanging curtains)(checked) have to be double-sided (checked) must have the same material on both sides (checked) not be sub-divided (checked)

·

the Full detailed calculation algorithm Late part ray-trace option must be used. If the standard RTC is selected it will automatically be changed before calculation. The reason is that with semi-transparent surfaces the reflection density growth is unknown and cannot be extrapolated.

87


·

In PLINFO and REFLINFO PLT-files, transmission coefficients are shown under the absorption coefficients in green and in the graph by a dotted green line.

·

it isn't obvious how to set the two coefficients (α/τ), see Theory for a details regarding the implementation of semi-transparency and how the coefficients can be estimated.

·

Sabine and other classical measures loose their meaning when semi-transparent materials are used since coupled room systems may be created, so be careful not to over-interpret them.

·

diffusion properties are controlled globally in the General settings dialog by the (Diffuse reflection) settings. Materials that do not have any diffusing properties assigned can get default values by selecting Surface default in the dialog.

(multple)

ABS1 …

Same as ABS above but all values are given in the range ]0..1[ instead of in %. NOROOM

This directive is used only in the master-file and when the GEO-file is not a room but a set of planes and markers to be exported using File|Export. With this directive, no source and receiver-file are required and some Geometry view/check plot-files are disabled. Note that even if a file is created only for export it has to contain at least one plane (but this plane does not have to be exported). Often a ground plane at z = 0 is useful as a reference plane to place e.g. a chair object on. Default: GEO-file is a normal room file NOHIDELINES

This directive is used when any hidden edges wants to be shown again temporarily e.g. for debugging purposes. Edges can be hidden by preceding corner ids in plane definitions by a minus-sign, see Planes section. Note that it is not often necessary to use this option since in the two main plot-files for geometry debug (PLINFO.PLT and COINFO.PLT) hidden edges are always shown but dashed. Default: Hidden edges are not shown

88


Corners section This section lists the corner/node ids and their x-, y-, and z-values. These corners can be used to create both planes and markers and be referred to in source- and receiver-files (using the x(), y(), z() functions).

CORNERS c_id c_x c_y c_z ...

The string CORNERS marks start of corner coordinates.

c_id:

a unique corner id (0-65535). See also OFFSETCO. Note: duplicate corner ids are always checked if any prediction is selected and if duplicates are found prediction is blocked.

c_x, c_y, c_z:

the coordinates as floating point numbers, declared constants, expressions or special functions as described below.

The corners can be entered in any order and corners not used don't have to be removed. Several tools for defining an accurate and structured geometry are provided. Use of Expressions:

Instead of first defining a symbolic constant (by using an expression) and then use the constant for a coordinate, the expression can be entered directly. The expression must, however, either be entered without intervening spaces or be completely enclosed in parenthesis. Without this convention it is not possible for the parser to know where, for example, the x-expression stops and the y-expression starts. Method no. 1 (slightly more efficient if x1, y1 and z1 are used many times):

LOCAL a = 5 LOCAL x1 = sqrt(a) LOCAL y1 = 3 + a LOCAL z1 = 1 - 2*a CORNERS 22 x1 y1 z1 ... PLANES ... Method no. 2 (more clear):

LOCAL a = 5 CORNERS 22 sqrt(a) ... PLANES ...

3+a

1-2*a ;without spaces parenthesis not required

or

LOCAL a = 5 CORNERS

89


22 (sqrt( a )) (3 + a) (1 - 2*a) ;with spaces and parenthesis ... PLANES ... Note: to avoid possible complications, an exception to the above is that an expression cannot start with a factor within parenthesis, i.e. an expression such as

(a+b)*sin(r) must either be completely enclosed within parenthesis:

((a+b)*sin(r)) or rearranged:

sin(r)*(a+b). The reason for this is that the parser looks for matched parenthesis if the first character is a "(" and will in the example stop after (a+b) and start the next coordinate with a "*" resulting in an error-message. Available functions :

sin(angle)

angle in degrees

cos(angle)

angle in degrees

tan(angle)

angle in degrees

arccos(value)

return value in degrees

arcsin(value)


arctan(value)


exp10(value)

exponential (base 10)

log10(value)

logarithm (base 10)

exp(value)

natural exponential (base e)

log(value)

natural logarithm (base e)

sqrt(value)

square root

sqr(value)

square: ()

cube(value)

cube: ()

abs(value)

absolute value

rnd(range)

returns a random number [0..range[.

rnd2(range)

returns a random number [-range..range[.

2

3

Of course, the rnd() and rnd2() functions are not very useful for geometry building but have been requested by a few users for randomly placing e.g. markers. An example of its use:

LOCAL actor_x = 3+rnd2(1) ; will let actor_z vary between 2 and 4) DEBUG actor_x ; displays the actual value Note: when the geometry is read, a checksum is created to ensure that if a Source addition is made from calculations at separate times they will be based on the same geometry. In such cases the rnd-functions should not be used for building the actual geometry.

90

2. PREDICTION MODULE Automatic locking to planes:

Locking hard-to-measure coordinates (e.g. a slightly sloping floor) into a plane can be accomplished by a function replacing either the x-, y-, or z-coordinate:

lock(id1 id2 id3) Where id1, id2 and id3 are the ids of three previously defined corners, describing the plane to lock into. An example:

CORNERS 1 ... 3 ... 15 ... 23 2.3 ... PLANES ...

15.4

lock(1 3 15)

where the z-coordinate of corner 23 will be calculated to lock into the plane defined by corners 1, 3 and 15 at x = 2.3 and y = 15.4 Plane-line intersection:

A corner can be completely defined by describing it to be the intersection of a line, described by two previously defined corner ids, and a plane described by three previously defined corner ids:

cut(l_id1 l_id2

p_id1 p_id2 p_id3)

where the first two ids describe the line and the last three describe the plane. An example:

CORNERS 1 ... 3 ... 15 ... 10 ... 11 ... 23 cut(10 11 ... PLANES ...

1 3 15)

where corner 23 will be calculated as the intersection of the line 10 to 11 with the plane described by corner 1, 3, and 15. Use of previously defined corners:

In addition to mathematical functions, functions returning the x-, y-, or z-value of a previously defined corner id can be used:

x(id) or y(id) or z(id) where id is a previously defined corner id (it must not be a symbolic constant or expression). An example:

CORNERS 12 2.2 23 2.3

13.4 15.4

13.2 z(12)

91


24 -x(23) ... PLANES ...

y(23)

z(12)+1

These three functions can be treated as any other functions, for example be used in expressions defining a constant. Absolute corner ids mixed with relative ids:

The convenient possibility to use relative corner ids in a file by specifying the OFFSETCO directive is sometimes complicated by the need to use a corner defined outside of the current GEO-file. To enable this, an id can be preceded by an underscore "_" and the id will not be incremented by the specified offset, and will thus be taken as an absolute id. An example:

OFFSETCO 100 CORNERS 12 1.2 2.5 23 2.3 y(_12) ...

3.3 z(12)

The y-value of corner 23 will be that of corner 12, while the z-value will be that of corner 112. Note: no space is allowed between the underscore and the id. A loop() function to generate coordinates:

The loop()function is used instead of an id x y z sequence, the syntax is:

loop(id_start,i,i_start,i_stop,i_step,x_expr,y_expr,z_expr) id_start

Start of corner ids to be generated by the function, for each new coordinate this number is incremented by 1

i

Loop variable

i_start

Integer start value for i

i_stop

Integer stop value for i

i_step

Integer step for i, can be < 0 for backwards stepping

x_expr

Expression for the x-coordinate (optionally including loop variable i )

y_expr

Expression for the y-coordinate (optionally including loop variable i )

x_expr

Expression for the z-coordinate (optionally including loop variable i )

Example creating coordinates in a crude circle (points every 45°): this sequence:

500 501 502 503 504 505 506

2*cos(0*45) 2*cos(1*45) 2*cos(2*45) 2*cos(3*45) 2*cos(4*45) 2*cos(5*45) 2*cos(6*45)

2*sin(0*45) 2*sin(1*45) 2*sin(2*45) 2*sin(3*45) 2*sin(4*45) 2*sin(5*45) 2*sin(6*45)

3 3 3 3 3 3 3

92


507 2*cos(7*45) 2*sin(7*45) 3 can be generated in one step by:

loop(500,angle,0,315,45,2*cos(angle),2*sin(angle),3) or by :

loop(500,q,0,7,1,2*cos(q*45),2*sin(q*45),3) Notes: ·

decimal comma cannot be used in the x-, y- and z- expressions but only decimal point (comma is used to separate parameters)

·

the selected loop variable must not clash with any other used letter in an expression (if sin() is used, s, i or n cannot be used as a variable)

·

the created coordinates can be mirrored, copied, translated etc. just like any other coordinates

·

if an error occurs the displayed loop() arguments will not look exactly as entered since all blanks have been removed (if lock() was used the blanks between the ids are replaced by a colon (:) i.e. lock(1 2 3) will become LOCK(1:2:3)). This is done to simplify parsing

Planes section PLANES [pl_id pl_name / c ... / abs ] or [pl_id pl_name / c ... / (sub_name / c ... / abs ) ...] ...

The second variant is for a sub-divided plane. The string PLANES marks start of plane definitions. [ ]

field delimiters: start and end of a plane definition (a plane definition can span several lines).

pl_id

a unique plane id number (1-99999, 0 allowed if OFFSETPL is used). Note: duplicate plane ids are always checked if any prediction is selected and if duplicates are found prediction is blocked.

pl_name

an (optional) name of the plane. The 30 first characters will be used (e.g. in PLINFO.PLT) the rest ignored. Note: slash "/" or backslash "\" can not be used in the name since they act as field delimiters.

/ \

field delimiter. Plane definitions in GEO-files can use either slash "/" or backslash "\" as field delimiters where "\" means that the plane corner entering order is reversed but only for the plane concerned. This feature has two uses: 1) to quickly correct a plane that by accident has been entered reversed, 2) to create doublesided planes by simply copying the definition of one side to a new line and flip / to \ and change plane id and absorption (if different on the two sides). An alternative is to use a special double-plane construction, see advanced options

93


below. Sub-divisions are automatically oriented in the same way as the plane and can always use e.g. /.

c ...

plane or sub-division corner ids entered in clockwise order as seen from the front or the back of the plane as specified by FROMFRONT|FROMBACK (from back is default). There is no practical limit of the number of corners in a plane or subdivision. Any corner will do as a start corner as long as the main direction is correct. It is not only allowed but strongly recommended to enter redundant corners on a plane (such as three points on a line) to use e.g. to connect to a neighbor plane. Planes cannot contain holes. See Fig 2.2 and 2.4. The mixing of absolute and relative corner ids, as described for the corners section, also applies in the planes section.

3

13 If FROMBACK is used, plane no. 10 is defined as:

plane no. 1

[10 / 11 12 13 14 / wood] or [10 \ 14 13 12 11 \ wood]

12

2

4

14

plane no. 10

1

If FROMFRONT is used, plane no. 1 is defined as:

[1 / 1 2 3 4 / wood] or [1 \ 4 3 2 1 \ wood]

11 Fig. 2.2 Plane corner entering order

abs

absorption/diffusion property name as described above for the ABS directive. If a material is used for one plane only, the values can be entered directly in the plane definition (e.g. [1 / 1 2 3 4 / <8 7 6 5 4 3> ]). If an asterisk "*" is appended to the name (or the values if no name is used) the plane or subdivision will have automatic edge diffusion (if also switched on in General settings).

( )

field delimiters: start and end of a plane sub-division definition. During the calculation, the plane sub-divisions are searched for hits in the order they are defined. This can be used to define one sub-division on top of another (max one layer). For example, a window and a door in a wall can be defined as the first and second divisions and the whole wall as the third division (the full wall is always the last division). If the window or the door is hit the search stops. If they are not hit then the hit must be in the wall outside the window and the door. This makes the plot of plane sub-divisions look more clear, saves time entering, and also uses less memory. See Fig 2.3. Note: sub-divisions cannot be created in the AutoCAD interface. Note: a sub-division cannot have semi-transparency, a double-sided plane with no sub-divisions has to be used.

sub_name a sub-division name (a name of any length can be used, but only the first 10 characters will be kept and shown in PLINFO.PLT). Planes can be entered in any order. There is a difference in how the plane definitions are entered as compared to all other data in the input-files which are entered and read line by line. To allow a definition to spread over many lines (e.g. if a plane has many corners or sub-divisions) characters are read until the first "[" is found and the definition is parsed until a matching "]" is found, see Fig 2.3a-c.

94


1 10

b 16

a 12 13

11 c

17

d

e

18 19

[1 2 [2 [3 [4 14 15 [5 [6 [7 f 20

a b c d e f g

/ 1 /16 /17 /18 /19 /20 / 4

21

15 11 12 13 14 15 17

14 13 12 11 10/ abs1] 17/ abs1] 18/ abs2] 19/ abs1] 20/ abs3] 21/ abs1] 18 19 20 21 3/ abs1]

The two windows (c with abs2, e with abs3) and the remaining wall (with abs1) modeled as separate planes (not recommended):

g 4

2 10 11 12 13 14 16

3

Fig. 2.3a Use of separate planes (not recommended)

1 10

a 12 13

11 b

16

[1 / 1 (a / 1 (b /16 (c /17 14 15 (d /18 (e /19 f (f /20 (g / 4 20 21 ] 2

c

17

d

e

18 19

2 15 21 3 4 16 10/ 2 15 14 13 12 11 10/ abs1) 10 11 17/ abs1) 11 12 18/ abs2) 12 13 19/ abs1) 13 14 20/ abs3) 14 15 21/ abs1) 16 17 18 19 20 21 3/ abs1)

g 4

Modeled with side-by-side subdivisions (not practical):

3

Fig. 2.3b Sub-optimal use of plane sub-divisions and with some risk of ray leaks

1

2 11

12 13 a

17

[1 /1 2 3 4/ (a /17 11 12 18/ abs2) (b /19 13 14 20/ abs3) (c / 1 2 3 4/ abs1) ]

14 b

18 19

Best modeled with a full wall subdivision followed by the windows (the order matters, full wall sub-division must be placed last)

20

c 4

3 Fig. 2.3c Optimal use of plane sub-divisions

Sometimes in modeling one has to resort to divide into triangles and many types of imports such as from AutoCAD will contain triangles. In other cases, such as a hole in a wall, division cannot be avoided. With divided planes the model will have many lines that are not actual plane edges and will be visually confusing. An option is to turn off the drawing of individual edges and is done by preceding a corner id with a minus-sign: [ 10 audience part 1 / 14 15 -39 / aud ] [ 11 audience part 2 / 39 -15 72 / aud ] where the edge for 15 to 39 in plane 10, and the edge 39 to 15 in plane 11 will not be drawn except in COINFO.PLT and PLINFO.PLT where they are drawn dashed. Another option is to use File|Export Geometry To|Merged Planes GEO-files.

95


Notes: ·

if an external corner id is referenced the minus-sign is given after the underscore, like this: _-39.

·

if the plane corner entering order is changed from the default (FROMBACK and //), the has to be placed on the other corner like in: [ 11 audience part 2 \ -39 15 72 \ aud ]

·

this is purely visual and does not in any way affect the calculations. A better option is to actually merge planes divided into unnecessary triangles into a new plane but that approach is likely to take longer time.

·

the directive NOHIDELINES can be used to to temporarily show all edges.

Advanced plane creation option 1 : id ranges in plane definitions

To speed up the creation of some planes and/or sub-divisions, corner ids can be given as ranges from-to: [31 ceiling / 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 / wood ] Can be defined with ranges, e.g.: [31 ceiling / 20-39 / wood ] [31 ceiling / 20 21 22-36 37-39 / wood ] [31 ceiling \ 39-20 \ wood ] ;ranges can also go from high to low Ranges can be combined with the hidden-edge codes: [32 floor / -1 -2 -3 -4 / wood ] [32 floor / -1-4 / wood ] as well as external ids: [33 wall / _11-15 / wood ] ;all ids are assumed external [33 wall / _-11-15 / wood ] ;same as above but with hidden edges

Note: no spaces are allowed within a range. Advanced plane creation option 2 : loops

To speed up the creation of some planes with logic numbering (say for circular structures) a loop-like construction for planes can be used: These five planes: [ [ [ [ [

1 2 3 4 5

wall wall wall wall wall

/ / / / /

1 2 3 4 5

2 3 4 5 6

3 4 5 6 7

4 5 6 7 8

/ / / / /

wood wood wood wood wood

] ] ] ] ]

can be created by one statement [ *5 1 wall / 1 1 1 1 / 1 2 3 4 / wood ]

96


Where *5 means that five planes are created incrementing the given plane id for each new plane, and the / 1 1 1 1 sequence means that each of the following corner ids will be increment by 1 for each new plane. Note: the * and the number of planes to create must not be separated by space. If the corner id increment is 1 for all corners (a common case), no increments needs to be given but just the extra /: [ *5 1 wall //

1 2 3 4 / wood ]

A case where different increments will be needed is e.g. with a cone-shape that has a common point 0 at the top: [ [ [ [ [

1 2 3 4 5

side side side side side

/ / / / /

0 0 0 0 0

1 2 3 4 5

2 3 4 5 6

/ / / / /

wood wood wood wood wood

] ] ] ] ]

and it can then be created by: [ *5 1 side / 0 1 1 / 0 1 2 / wood ] With sub-divisions the same principle holds and the corner id increments are given the same way, e.g. as: [ *19 1 / 1 1 1 1 / 0 1 21 20 / (window / 1 1 1 1 / 100 101 121 120 / wnd) (wall / 1 1 1 1 / 0 1 21 20 / steel) ] or using shortcuts for increment 1: [ *19 1 // 0 1 21 20 / (window // 100 101 121 120/ wnd) (wall // 0 1 21 20 / steel ) ]

An example where different increments than 1 may be needed: 1

11

2

3

5

4

20

21

24

25

28

29

32

33

23

22

27

26

31

30

35

34

12

13

Fig. 2.4 Sample plane-loop use

[ *4 1 / 1 1 1 1 / 1 2 12 11 / (window / 4 4 4 4 / 20 21 22 23 / glass) (wall / 1 1 1 1 / 1 2 12 11 / wood) ]

or with the shortcut when the increment is 1:

97

14

15


[ *4 1 // 1 2 12 11 / (window / 4 4 4 4 / 20 21 22 23 / glass) (wall // 1 2 12 11 / wood) ]

Another example where e.g. an increment of 2 may be needed is if in a circular structure every second plane has a different absorption. The corner id increments given for plane loops can also be negative. These two will give the same result: [*3 25 refl front / 1 1 1 1 / 0 1 2 3 / frontabs* ] or just [*3 25 refl front // 0 1 2 3 / frontabs* ]

giving corner ids 0 1 2 3 1 2 3 4 2 3 4 5

and [*3 25 refl front / -1 -1 -1 -1 / 2 3 4 5 / frontabs* ]

instead giving ids 2 3 4 5 1 2 3 4 0 1 2 3

Planes created in a loop will have " (n)" appended to the plane name where n is the loop index. I.e. if the plane name given is "wall", the names of loop-created planes will be "wall", "wall (2)", "wall (3)" etc. Advanced plane creation option 3 : double-sided planes

The creation of double-sided planes (with or without sub-divisions) can be made simplified: [ 1 refl / 1 2 3 4 / wood ] [ 2 refl \ 1 2 3 4 \ wood ]

can be created by one statement by putting a "D" or "d" before the plane id: [ D 1 refl / 1 2 3 4 / wood ]

and the "other" side will also have " - back" appended to the plane name. The speed-up is not dramatic but there is also some benefit in clearly seeing that it is a double plane in the GEO file and that it may be less likely that the back side is forgotten. This also works together with plane loops (D is always placed after *n) , e.g. as in: [ *5 D 1 wall // 1 2 3 4 / wood ]

Where five double planes are created incrementing the given plane id for each new plane (every second created plane will be the back side). A typical use here would be a range of ceiling reflectors (although then COPY should be better seeing each double-sided reflector as an OBJECT). Note: this function is only for double-sided planes that have the same material on both sides. If

98


the materials differ it will have to be done by one plane definition per side.

Markers section Markers are non-acoustical elements not affecting the prediction but are for visual appearances only. Markers can also be used to create sound source Visuals. Marker-files can indirectly be made in AutoCAD via the AutoLISP interface and then modified in GEO-format and exported as a marker-file. A directive NOROOM can be used if a GEO-file is used only to create a Marker-file (no source and receiver files necessary, no calculations made). All markers except LOAD have some common initial parameters: NAME ids... {R G B} offset

ids...

a number of corner ids used to determine the location, direction and drawing plane of the marker object. Even for two-point lines a drawing plane needs to be defined so that the direction of the drawing offset can be determined.

{R G B}

the Red Green Blue values for the marker color, each in the range [0..255] e.g. {255 0 0} for pure red. CATT-Edit has a Template| RGB Color menu item that inserts an RGB-triplet from visual selection in a color dialog.

offset

an offset from the plane defined by the ids. The idea is to always offset the marker against e.g. a wall so that with shading, the marker is shown in front of the wall. With loudspeaker visuals the offset is typically set to zero except for e.g. outer markings such as a brand label.

POLY id1 id2 ... idN {R G B} offset

A filled planar (i.e. all points are in the same plane) polygon marker with no outlined edge: id1 id2 ... idN the corner ids defining the polygon. The polygon is automatically closed (i.e. the first id does not have to be entered also as the last id). The polygon normal is calculated from the first three ids in the following way: v12 = vector(id1→id2), v23 = vector(id2→id3) and normal_vector = cross_product(v12,v23). This corresponds to entering the ids in anticlockwise order as seen from the desired front side. {R G B}

are the Red Green Blue values for the fill color [0..255]. Only the front side of the polygon is filled (i.e. the side in direction of the normal).

offset

is the offset (in meter) from the plane (defined by id1, id2 and id3), and in direction of the normal, to where the polygon will be drawn.

Example: POLY 1 2 3 4 5 6 7 {255 0 0} 0.1

i.e. a red polygon with 7 corners drawn 0.1 m in front of the plane defined by ids 1, 2 and 3. Tip: To create an outline use the LINE marker with same ids but repeat id1 at the end: LINE 1 2 3 4 5 6 7 1 {0 0 0} 0.1

i.e. a black outline.

99

2. PREDICTION MODULE LINE id1 id2 ... idN {R G B} offset pattern

A multi-point planar (i.e. all points are in the same plane) line marker: id1 id2 ... idN the corner ids defining the line. To make a closed line idN has to be the same as id1. The line normal is calculated from the first three ids in the following way: v12 = vector(id1→id2), v23 = vector(id2→id3) and normal_vector = cross_product(v12,v23). This corresponds to entering the ids in anti-clockwise order as seen from the desired front side. {R G B}

are the Red Green Blue values for the line color [0..255].

offset

is the offset (in meter) from the plane (defined by id1, id2 and id3) and in direction of the normal, to where the line will be drawn.

pattern

optional pattern : 0 (solid), 1, 2, 3, 4. Default is a solid line.

Examples: LINE 1 2 3 4 5 6 7 {255 0 0} 0.1

i.e. a seven-point red line with drawn 0.1 m in front of the plane defined by ids 1, 2 and 3. DASH id1 id2 id3 {R G B} offset pattern

A two-point line marker: id1 id2 id3

the actual line is between id1 and id2. The line normal is calculated from the ids in the following way: v12 = vector(id1→id2), v23 = vector(id2→id3) and normal_vector = cross_product(v12,v23). This corresponds to entering the ids in anti-clockwise order as seen from the desired front side.

{R G B}

are the Red Green Blue values for the line color [0..255].

offset

is the offset (in meter) from the plane defined by the ids, and in direction of the normal, to where the dash will be drawn

pattern


Examples: DASH 1 2 3 {255 0 0} 0.1

i.e. an red dash with drawn 0.1 m in front of the plane defined by ids 1, 2 and 3. RULE id1 id2 id3 {R G B} offset tic pattern

A two-point ruler marker with tics: id1 id1 id3

the actual ruler line is between id1 and id2. The line normal is calculated from the ids in the following way: v12 = vector(id1→id2), v23 = vector(id2→id3) and normal_vector = cross_product(v12,v23). This corresponds to entering the ids in anti-clockwise order as seen from the desired front side.

{R G B}

are the Red Green Blue values for the ruler color [0..255]. Tics will be drawn in a complementing color.

100


offset

is the offset (in meter) from the plane defined by the ids, and in direction of the normal, to where the ruler will be drawn

tic

the tic size (in meter). Every fifth tic will be drawn in a different color.

pattern


Example: RULE 1 2 3 {255 0 0} 0.1 1.0

i.e. an red ruler between 1 and 2 drawn 0.1 m in front of the plane defined by ids 1, 2 and 3. DISC id1 id2 id3 {R G B} offset radius

A filled disc marker: id1 id1 id3

id1 is the disc center. The disc normal is calculated from the ids in the following way: v12 = vector(id1→id2), v23 = vector(id2→id3) and normal_vector = cross_product(v12,v23). This corresponds to entering the ids in anti-clockwise order as seen from the desired front side.

{R G B}

are the Red Green Blue values for the disc fill color [0..255].

offset

is the offset (in meter) from the plane defined by the ids, and in direction of the normal, to where the circle will be drawn

radius

disc radius (in meter)

Example: DISC 1 2 3 {255 0 0} 0.1 1.0

i.e. a 1.0 m radius red disc drawn 0.1 m in front of the plane defined by ids 1, 2 and 3. CIRC id1 id2 id3 {R G B} offset radius pattern

A circle marker: id1 id1 id3

id1 is the circle center. The circle normal is calculated from the ids in the following way: v12 = vector(id1→id2), v23 = vector(id2→id3) and normal_vector = cross_product(v12,v23). This corresponds to entering the ids in anti-clockwise order as seen from the desired front side.

{R G B}

are the Red Green Blue values for the circle line color [0..255].

offset

is the offset (in meter) from the plane defined by the ids, and in direction of the normal, to where the circle will be drawn

radius

circle radius (in meter)

pattern

optional pattern : 0 (solid), 1, 2 3, 4. Default is a solid circle.

Example: CIRC 1 2 3 {255 0 0} 0.1 1.0

101


i.e. a 1.0 m radius red circle drawn 0.1 m in front of the plane defined by ids 1, 2 and 3. TEXT id1 id2 id3 {R G B} offset "text" height width style

A 3D text marker: id1 id1 id3

the text is drawn from id1 along vector(id1→id2) and with up defined by vector(id2→id3) (this vector does not have to be perpendicular to the id1→id2 vector). The text normal is calculated from the ids in the following way: v12 = vector(id1→id2), v23 = vector(id2→id3) and normal_vector = cross_product(v12,v23). This corresponds to entering the ids in anti-clockwise order as seen from the desired front side.

{R G B}

are the Red Green Blue values for the text color [0..255].

offset

is the offset (in meter) from the plane defined by the ids, and in direction of the normal, to where the text will be drawn.

text

the actual text (entered within " ")

height

the text height in meter, measured from the baseline to the top of capitals.

width

the average text width in meter (the text is not a constant width text)

style

optional style : B is bold (no other styles implemented)

Example: TEXT 1 2 3 {255 0 0} 0.1 "Wall" 1.0 0.5

i.e. the red text Wall with 1.0 m high and 0.5 m wide characters drawn from id1 in the direction 1→2 with 2→3 defined as up. LOAD

id1 id2 id3 "path\mrk-file"

A marker that loads a marker-file object (.MRK). Marker-file objects can be created in GEOformat and exported to file using the File|Export menu: id1 id1 id3

the marker is loaded at id1 with its x-axis along vector(id1→id2) with vector(id1→id3) defined as up (this vector does not have to be perpendicular to the id1→id2 vector). id3 is optional and if not given the marker is placed assuming up is the z-axis. Note that any offsets should be made already when creating the marker object.

path\mrk-file the marker-file to load, if path is not given the file is assumed to be located in the project input folder. The .MRK extension is optional.

Example: LOAD 1 2 3 "Chair"

i.e. the Chair.MRK marker-file object will be loaded at corner 1 with its x-axis defined by 1→2 and z-axis by 1→3 (the y-axis is calculated).

102


Modeling notes Planes must not be entered on top of each other (use plane sub-divisions instead as explained above). If a triangle-plane has all its corners in a line the plane is removed and an error message is given. The cause of this error may either be a mistake (that has to be corrected) or as a result of an export from AutoCAD that occasionally has been found to give these types of nonplanes. Corners/nodes that touch edges of adjacent planes should be included in both planes, see Fig. 2.5: 2 13

12 11 1

3

14

[1 big / 1 14 13 2 3 4 / abs1 ] [2 small / 11 12 13 14 / abs2 ] 4

Fig. 2.5 Attaching plane corners properly

If corners 13 and 14 are not included in plane 1, debug will generate an "Edges cutting or touching error" and also that 13 and 14 are "single-connected". These errors are not severe but should, if possible, be avoided. When modeling from scratch in the GEO-format it is not difficult but models imported from AutoCAD will typically have a lot of these errors. For volume calculations to be correct models must be made without single-sided overlapping planes. It can be described like this: "all planes that do not have their backside towards the outside of the room must be made double-sided". and “all planes that can be reached by sound must be modeled - but only those planes must be modeled and only those parts that can be reached by sound” The same rules holds for the "potentially reversed normal" debug check to work. The debug option "single-connected corners" flags some problem planes but not all (e.g. the underside of an audience must be included even if the audience is "boxed-in" if the floor under the audience is still modeled, see example below). Another way to describe the matter is just like the volume calculation algorithm works: "if a ray from any position inside a room, sent out in any direction, passes an odd number of planes on its way to infinity the position is inside, otherwise it is outside". From this it can be seen that if one audience plane is placed on top of a full floor there can be two passings of a ray signaling a truly inside position to be outside. Any automatic calculation from an arbitrary geometry needs to rely on that the geometry is built according to certain rules. Figures below only show the floor/audience part but it is assumed that walls and ceiling are present too.

103


P audience surface 1 Q

1

floor

2

P: outside room (false) Q: inside room (correct) Fig. 2.6 Geometrically incorrect method where volume estimate, surface area and classical mean absorption will be wrong (Sabine RT). Rays will also pass under and through the audience (from below).

P 1 Q

1

2

3

P: inside room (correct) Q: inside room (correct) Fig. 2.7 Geometrically correct method but volume estimate, surface area and classical mean absorption will be wrong (Sabine RT). Rays will also pass under the audience.

P 1 Q

1

3

audience edges 2

P: inside room (correct) Q: inside room (false) Fig. 2.8 Geometrically correct method but volume estimate, surface area and classical mean absorption will be wrong (Sabine RT).

This is the correct way and the audience block will be excluded from the volume: P 1 Q

P: inside room (correct) Q: outside room (correct) Fig. 2.9 Geometrically correct method including correct volume and Sabine RT.

104


Example model Below are the geometry-files for a simple hall. These files are in the CATTDATA-folder MODELS\EXAMPLE. Constants and INCLUDEs are used a lot to demonstrate the principle (and power) of these two concepts. Main symbolic constants used can be seen in Fig 2.10 and main model corners in Fig. 2.11.

Fig. 2.10 Description of constants used in the example hall.

108 300

106

8

301 303

104 203 107 213 6

101

202 2127

4

103

204 105214 102

01

3 A0

2 1

302

201 211 5

Fig. 2.11 COINFO.PLT for the example hall.

105


The GEO-file inclusion "tree" looks like this:

MASTER.GEO SIDES.GEO ACROSS.GEO AUDIENCE.GEO REFL.GEO MARKERS.GEO The GEO-files: GEO-file MASTER.GEO ;CATT-Acoustic v8.0 Manual Example Hall

; used in REFL.GEO to add more reflectors (if set to TRUE) GLOBAL showcopy = FALSE INCLUDE INCLUDE INCLUDE INCLUDE INCLUDE

sides.geo across.geo audience.geo refl.geo markers.geo ;some non-acoustic visual elements

ABS birch = < ABS somewood = ABS absofix = ; 8k and 16kHz ABS absorber = ABS custom = < ABS diffusor =

20 15 12 10 7 5 > L < 10 12 14 18 20 20> {159 152 96} birch < 20 30 40 50 60 70 : 90 95> {94 119 162} added absofix 20 15 12 10 7 5 > L < 20 30 40 50 60 70 > {228 179 27} custom

; general dimensions - change theses and the rest of the hall will ;follow ! ; If you like to experiment change these to GETGLOBAL. GLOBAL sh = 6 ; stage wall height GLOBAL sd = 9 ; stage depth GLOBAL sw = 11/2 ; stage back wall width/2 GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL

ph pw pth pbh psw

= 10 = 17/2 = 3 = 1.2 = 4.5

GLOBAL hd = 30

; ; ; ; ;

prosc prosc prosc prosc prosc

height width/2 top wall height bottom wall height side wall width (1.5)

; hall depth (without stage)

GLOBAL bww = 14/2 ; back wall width/2 GLOBAL bwh = 10 ; back wall height GLOBAL fltilt = 2 ; floor tilt CORNERS ; section is empty PLANES ; section is empty

106


GEO-file SIDES.GEO ;CATT-Acoustic v8.0 Manual Example Hall

OFFSETPL 10 MIRROR 100 10 CORNERS 1 sw -sd 0 2 pw 0 0 3 sw -sd sh 4 pw 0 ph 5 pw+psw 0 -pbh 6 x(5) 0 ph+pth 7 bww hd fltilt-pbh 8 bww hd z(7)+bwh ; MIRROR creates 101 - 108 PLANES [0 stage wall / 1 [1 prosc edge / 2 [2 side wall / 5 ;creates planes 10

3 4 6 -

4 2 / somewood ] 6 5 / absorber ] 8 7 / somewood ] 12 and MIRROR creates 20 - 22 (OBS OFFSETPL 10)

GEO-file ACROSS.GEO ;CATT-Acoustic v8.0 Manual Example Hall

LOCAL ay1 = 6 LOCAL ay2 = hd - 1 CORNERS 401 402 403 404

-ay1*x(5)/hd -ay2*bww/hd -x(402) -x(401)

ay1 ay2 y(402) y(401)

lock(106 lock(106 lock(106 lock(106

108 108 108 108

6) 6) 6) 6)

PLANES [30 stage wall / 101 103 3 1 / somewood [31 stage ceil / 103 104 4 3 / somewood [32 stage floor / 1 2 102 101 / somewood [33 bottom prosc / 105 102 2 5 / absorber [34 top prosc / 104 106 6 4 / absorber [35 hall ceiling / 106 108 8 6 / ( d / 401 402 403 404 / diffusor) ( w / 106 108 8 6 / somewood) ] [36 back wall / 7 8 108 107 / somewood ] GEO-file AUDIENCE.GEO ;AUDIENCE.GEO ;CATT-Acoustic v8.0 Manual Example Hall

OFFSETCO 200 OFFSETPL 50

107

] ] ] ] ]


ABS aud = <35 53 73 80 79 75 : 80 90> L <30 40 50 60 70 70> {153 214 41} (on the same line) LOCAL delta = 1.2 LOCAL audh = 1.2 LOCAL LOCAL LOCAL LOCAL LOCAL LOCAL

x1 x2 y1 y2 z1 z2

= pw + psw - delta = bww - delta = delta = hd - delta = -pbh + audh = z1 + fltilt

;string constants used in plane names below LOCAL a = "audience" LOCAL f = "floor" CORNERS ;audience upper corners 1 x1 y1 z1 ; corner 1 could as well be written with the equations like: ;1 pw+psw-delta delta -pbh+audh ;or ;1 (pw + psw - delta) delta (-pbh + audh) 2 x2 y2 z2 3 -x2 y2 z2 4 -x1 y1 z1 ;audience ;note use applied) 11 x1 y1 12 x2 y2 13 -x2 y2 14 -x1 y1

floor corners of _ to access absolute external id:s (OFFSETCO is not lock(_5 lock(_5 lock(_5 lock(_5

_105 _105 _105 _105

_7) _7) _7) _7)

PLANES [0 [1 [2 [3 [4 [5 [6 [7 [8

%a% / 1 2 3 4 / aud ] %a% front / 1 4 14 11 %a% back / 12 13 3 2 %a% side / 11 12 2 1 %a% side / 13 14 4 3 front %f% / _5 11 14 back %f% / _7 107 13 side %f% / _5 _7 12 side %f% / 107 105 14

/ aud / aud / aud / aud 105 / 12 / 11 / 13 /

] ] ] ] somewood somewood somewood somewood

GEO-file REFL.GEO ;CATT-Acoustic v8.0 Manual Example Hall

LOCAL LOCAL LOCAL LOCAL

w = bww - 1.8; 0.2 d = 7 ty = 24.3 tz = 10.5

108

] ] ] ]


OBJECT ROTATE -40 0 0 TRANSLATE 0 ty tz ;set showcopy = TRUE i MASTER.GEO to add these! IF showcopy THEN COPY 5 2 0 0 0 0 ty-1*(d+0.1) tz 0 0 0 COPY 10 4 0 0 0 0 ty-2*(d+0.1) tz 0 0 0 COPY 15 6 0 0 0 0 ty-3*(d+0.1) tz 0 0 0 ENDIF OFFSETCO 300 CORNERS 0 -w 0 0 1 -w d 0 2 w d 0 3 w 0 0 PLANES ;* = size ;(must be [60 refl ;the back [61 refl

dependent auto edge diffusion enabled in General settings) front / 0 1 2 3 / somewood* ] of a reflector MUST be defined too back \ 0 1 2 3 \ absorber* ]

GEO-file MARKERS.GEO ;CATT-Acoustic v8.0 Manual Example Hall ;Places a wall clock above the stage and adds a ruler to ;the stage wall

LOCAL clockradius = 0.55 LOCAL minhand = 0.5 ; length of clock minute hand LOCAL hourhand = minhand*0.8 ; length of clock hour hand CORNERS ;coordinates to place the "stage wall" text a bit over the floor 1001 x(1) y(1) z(1)+1 1002 x(2) y(2) z(2)+1 ;coordinates for loading the chair 1010 0 y(2)-2 z(2) 1011 -2 y(2)-2 z(2) ;coordinates to place the clock 1020 0 y(4) z(4)+2 ;clock center 1021 x(1020)+hourhand*cos(45) y(4)+0.01 ;hour hand end 1022 x(1020)+minhand*cos(90) y(4)+0.01 ;min hand end

z(1020)+hourhand*sin(45) z(1020)+minhand*sin(90)

PLANES ;empty MARKERS TEXT 4 104 106 {255 0 0} 0.1 "v8 Example Hall" 1.5 0.7 TEXT 1002 1001 3 {0 0 255} 0.1 " stage wall" 0.8 0.5

109


LOAD 1010 1011 "bluechair" ;clock DISC 1020 6 4 {0 255 0} 0.05 clockradius CIRC 1020 6 4 {0 255 255} 0.05 clockradius LINE 1021 1020 1022 { 0 0 0 } 0.12 ; the hands ;Ruler to show the stage height RULE 2 4 6 {0 0 0} 0.2 1.0

Source-file The source data has to be specified in a separate file with the extension .LOC in the input folder as specified in the General settings dialog. This file is read after the GEO-files and can therefore use global constants declared in these files. Normal name is SRC.LOC.

Syntax The principles are the same as stated for the GEO-files in the previous sub-section. A sourcefile must have two sections with the following order: directives (constant declarations etc.) SOURCEDEFS source locations, aim, octave-band levels and delay. The directives section may be empty but the string SOURCEDEFS must be present and at least one source must be defined.

Directives section SCALE sc_x sc_y sc_z



+ options

GETLOCAL name = expr|string

(multiple)

These have the same meaning as in GEO-files. There is no point in defining global constants in the source-file since it is the last file to be read.

Sourcedefs section With the prediction of measures that are not only relative early/late measures (SPL, STI and RASTI) the level at 1m in front of a source must be set to the actual level. However, for auralization the acoustical source spectrum (of a talker, singer or instrument) is present in the anechoic speech/music and is removed from calculated echograms before post-processing leaving only the system gain/eq to affect the sound (for natural sources the gain is 0 dB). This eq/gain can optionally be switched off at post-processing (as if the system was ideally flat with 0 gain). By removing the acoustical source spectrum it is also reasonable to switch anechoic material as long as the implied source has a directivity pattern similar to the one given in the source-file.

110

2. PREDICTION MODULE SOURCEDEFS If a source is of natural type (talker, singer, instrument, machine):

s_id s_x s_y s_z dirname[.SD0|.SD1|.SD2] a_x a_y a_z [roll [MAPRAYFACTOR = f]] Lp1m_a = Lp1m_a_description

(on one line)

or

s_id s_x s_y s_z anechoic.WAV a_x a_y a_z [roll [MAPRAYFACTOR = f]] Lp1m_a = Lp1m_a_description

or as one of the above where a_x a_y a_z is replaced by the aim(H,V) function. Where:

s_id:

source id (A0..A9 - Z0..Z9 allowing for 260 sources). The sources can be entered in any order and in any combination.

s_x, s_y, s_z:

source center location coordinate as floating point numbers, declared constants, or expressions. The lock(), cut(), x(), y(), z() functions can also be used (e.g. to lock a source to a fixed distance from a movable wall). Note: a source must not be placed exactly inside a wall since it will result in ray loss.

dirname

source directivity name in the CATTDATA folder SD (filename dirname.SD0, dirname.SD1, dirname.SD2, dirname.CF1 or dirname.CF2) or in a sub-folder to SD (filename subfolder\ dirname.SD0 etc.) or OMNI for a predefined omni-directional natural source. If no extension is specified SD0 is assumed. A source-name can be copied using the Directivity module Utilities menu and pasted into CATT-Edit while editing the source-file.

a_x, a_y, a_z:

coordinate of where to aim the source. The lock(), cut(), x(), y(), z() functions can also be used. When using the predefined OMNI source, the aim has no effect but must be given, it is useful for the 3D-viewer source view points.

aim(H,V):

source aiming angles as an alternative to an aim coordinate. Angle conventions according to fig. If aim() is used, an aim line is calculated to where the first surface is hit.

z

-180° ≤ H ≤ 180° -90° ≤ V ≤ 90° H=0°,V=0° is the y-axis H > 0° on the positive x side V > 0° on the negative z side

V

y

roll:

H optional roll angle, in degrees, around the source axis (defined by x s → a) looking down the axis from aim line s_to a_. Positive angles assume clockwise rotation around the axis. The roll angle is indicated in the SRCINFO.PLT geo-check plot-file.

f

(MAPRAYFACTOR) a factor 0.1..1.0 to apply to the number of rays chosen in audience area mapping with this source. If the map ray factor is used, also the roll angle must be given (even if zero). Note: If the factor given will make the number of rays used be less than 100 it will be selected as 100

111


(no matter the number of rays the direct sound is the same). If this option is used it is documented in the plot-file title bar as /F and is also indicated in MAPSTAT.TXT along each source listed

anechoic.WAV

means that the directivity is taken from the anechoic WAV-file info: CATTDATA-folder ANECHOIC\anechoic.AWI (if present for the anechoic file). See Post-Processing:Utilities for how to create AWI-files.

Lp1m_a_description:

< Lpa125 .. Lpa4k [ : Lpa8k Lpa16k ] > the SPL values at 1m on axis of the natural source. Lpa8k and Lpa16k are sound pressure levels for the 8 and 16 kHz octavebands. These values are optional if not included they are extrapolated from 2k and 4k values. Lp_nominal

the Nominal SPL 1m values on axis (as set in the Directivity module for the directivity)

Lp_max

the Max SPL 1m values on axis (as set in the Directivity module for the directivity)

Lp_voice_normal

shorthand for (predefined normal voice level constants, see below)

Lp_voice_raised

shorthand for (predefined raised voice level constants, see below)

Lp_voice_loud

shorthand for (predefined loud voice level constants, see below)

Lp_voice_IECmale LpA shorthand for an IEC 2003 60286-16 [IEC 2068-16] male speech spectrum with an LpA A-weighted level, to be used for STI. Relative 125-16k 1/1-octave spectrum (giving 0 dBA) is: 2.9 2.9 -0.8 -6.8 -12.8 -18.8 -24.8 (-35) dB Lp_voice_IECfemale LpA shorthand for an IEC 2003 60286-16 female speech spectrum with an LpA A-weighted level, to be used for STI. Relative 12516k 1/1-octave spectrum (giving 0 dBA) is: (-15) 5.3 -1.9 -9.1 -15.8 -16.7 -18 (-35) dB Lp_white Lpa1k

creates a white spectrum (+3dB/octave) with Lpa1k at 1kHz

Lp_pink Lpa1k

creates a pink spectrum (0dB/octave) with Lpa1k at 1kHz

anechoic.WAV

SPL values at 1 m on axis taken from an anechoic WAV-file info: CATTDATA-folder ANECHOIC\anechoic.AWI (if present for the anechoic file given). See Post-Processing:Utilities for how to create AWI-files.

If a source is of electro-acoustic type (loudspeaker of some kind):

s_id s_x s_y s_z directivity[.SD0|.SD1|.SD2|.CF1|.CF2] a_x a_y a_z [[roll] MAPRAYFACTOR = f] (on one line) Lp1m_a = Lp1m_a_description Lp1m_ea = Lp1m_ea_description Delay_e = delay

112


or

s_id s_x s_y s_z directivity[.SD0|.SD1|.SD2|.CF1|.CF2] a_x a_y a_z [[roll] MAPRAYFACTOR = f] (on one line) Lp1m_a = Lp1m_a_description Gain_a = < G125 .. G4k [ : G8k G16k ] > Delay_e = delay or as one of the above with a_x a_y a_z replaced with aim(H,V). Where: Lp1m_a_description:

< Lpa125 .. Lpa4k [ : Lpa8k Lpa16k ] > the SPL values at 1m on axis of the acoustical source (see Lp1m_a_description above about the 8k and 16k values) Lp_voice_normal

see above

Lp_voice_raised

see above

Lp_voice_loud

see above

Lp_voice_IECmale LpA see above Lp_voice_IECfemale LpA see above Lp_white Lpa1k

see above

Lp_pink Lpa1k

see above

Lp_sensitivity

uses the source Sensitivity and with Gain_a = < 0 0 ...> the SPL at 1m on axis (Lp1m_ea) will then roughly correspond to that of 1 W input (assuming 8 ohm). Note: CF1 and CF2 files do not always have Sensitivity specified (e.g. not for an active or powered loudspeaker). Note: when using this option with auralization and a flat Gain_a it will be as if the speaker was flat since Lp1m_a is removed before creating the impulse response and is instead replaced by the spectrum of the anechoic input WAV. If the eq effects of a non-flat Sensitivity is to affect the auralization, the corresponding eq values have to be added to Gain_a.

anechoic.WAV

see above.

Lp1m_ea_description:

< Lpea125 .. Lpea4k [ : Lpea8k Lpea16k ] > the SPL values at 1m on axis of the loudspeaker (see Lp1m_a_description above about the 8k and 16k values)

Lp_max

see above

Lp_white Lpea1k

see above

113


Lp_pink Lpea1k

see above

< G125 .. G4k [ : G8k G16k ] > is the system total gain (Lp1m_ea is then calculated to be = Lp1m_a + Gain_a) delay

delay in ms (has to be given even if it is zero). The delay is taken into account for Audience area mapping, Source addition and multiple source auralization using SIM-file and WAV-file addition.

For remaining parameters for an electro-acoustic source (s_id, s_x, f, etc.) see natural source. Examples assuming Talker.SD0 is a natural source and Horn.SD1 is a loudspeaker:

LOCAL g = 10 ;system gain SOURCEDEFS ;A0 and A1 give the same Lp1m_a levels expressed in two ways A0 -1 2 1.7 Talker.SD0 0 3 1.7 Lp1m_a = A1 1 2 1.7 Talker.SD0 0 3 1.7 Lp1m_a = Lp_voice_loud ;B0 and B1 give the same Lp1m_ea levels expressed in two ways B0 -1 -1 1.7 Horn.SD1 0 3 1.7 Lp1m_a = ; the spectrum of the talker Lp1m_ea = Delay_e = 0 ; delay has to be given even if it is 0 B1 1 -1 1.7 Horn.SD1 0 3 1.7 Lp1m_a = Lp_voice_normal Gain_a = Delay_e = 0 ;C2 uses the anechoic WAV-file info embedded in an AWI-file ; assumes that Mt_44_an.AWI has directivity and Lp 1m info. C2 1 -1 1.7 Mt_44_an.WAV 0 3 1.7 Lp1m_a = Mt_44_an.WAV Note: if SCALE is used in all GEO-files as well as in the receiver- and source-files using x(), y(), and z() functions will scale twice since scaling has already been performed. In such a case use e.g. 1+x(12)/s, where s is the scale factor, to prevent the extra scaling. Note: the predefined OMNI source is treated as an natural source, if an electro-acoustic source will be used with omni-directional directivity a directivity file has to be created for it with a name different from OMNI.SD0. Since many “omni” sources are not very omni at mid and high frequencies a measured directivity is preferred when comparing to measurements. Various octave-band speech spectrums at 1 m distance have been defined by ANSI S3.79 (draft) as spectral densities at the center frequencies and as an average for males and females. The corresponding Lp1m_a values have been calculated and are included as predefined global constants (as well as used in Lp_voice_normal etc.):

114


Normal vocal effort Raised vocal effort Loud voice N125 N250 N500 N1k N2k N4k N8k N16k

= = = = = = = =

51.2 57.2 59.8 53.5 48.8 43.8 38.8 33.8

R125 R250 R500 R1k R2k R4k R8k R16k

= = = = = = = =

55.5 61.5 65.6 62.4 56.8 51.3 42.5 33.8

L125 L250 L500 L1k L2k L4k L8k L16k

= = = = = = = =

58.0 64.0 70.3 70.7 65.9 59.9 48.9 37.8

(extrapolated from 4k and 8k)

Note: the 125 Hz SPL is not defined in the standard and has been set to -6dB lower than the 250Hz band. The 16k values are extrapolated from 4k and 8k. A useful information may be that the three levels of speech correspond to the following overall A-weighted values: 59.5 dBA (normal), 66.5 dBA (raised) and 73.6 dBA (loud). An alternative to using these speech levels are the Lp_voice_IEC... shortcuts.

Example The example below is the CATTDATA-folder\MODELS\EXAMPLE\SRC.LOC. See fig 2.9.

;CATT-Acoustic v8.0 Manual Example Hall LOCAL sh = 1.7 SOURCEDEFS A0 3.0 -2.0 sh OMNI 0.0 10.0 2.0 Lp1m_a = <70 73 76 79 82 85> B0 -2.0 -5.0 sh OMNI 0.0 10.0 2.0 Lp1m_a = <70 73 76 79 82 85>

Receiver-file The receiver data has to be specified in a separate file with the extension .LOC in the input folder as specified in the General settings dialog. Normal filename is REC.LOC.

Syntax The principles are the same as stated for the GEO-files. The receiver-file must have two sections with the following order : directives (constant declarations etc.) RECEIVERS receiver locations and optional individual head-directions The directives section may be empty but the string RECEIVERS must be present and at least one receiver must be defined.

115


Directives section SCALE sc_x sc_y sc_z GLOBAL name = expr|string #G name = expr|string



(multiple) (short form) (+ options) (+ options)

GETGLOBAL name = expr|string GETLOCAL name = expr|string

(multiple) (multiple)

These have the same meaning as in GEO-files. Global constants declared in the receiver-file will be known also in the source-file.

Receivers section RECEIVERS r_id r_x r_y r_z [h_x h_y h_z] [ < n125 n250 n500 n1k n2k n4k [ : n8k n16k ] ] > [OVERRIDE] (on one line) ...

The string RECEIVERS to mark start of receiver locations.

r_id:

receiver id (0-99). The receivers can be entered in any order and in any combination.

r_x, y, z:

receiver location coordinates as floating point numbers, declared constants, or expressions. The lock(), cut(), x(), y(), z() functions can also be used (e.g. to lock all receivers to a fixed distance above the audience planes). Note: a receiver must not be placed exactly inside a wall since it will result in ray loss.

h_x, y, z:

optional individual receiver head-direction. If not given, the selection in General settings is used (Stage, Source or Fixed). Useful e.g. with the walkthrough convolver since receiver positions and head-directions can be created along a path in a 3D graphics-software walkthrough.

n125...

As an alternative to using the constant background noise in General settings/Acoustic Environment (Env.) in STI and RASTI and U50 calculations, each receiver can be assigned a specific background noise (estimated from a noise map or via a separate Full detailed calculation with only noise sources). The STI and RASTI plot-files will indicate the constant background level but mark with the letter N all receivers that have individual noise (the information on the actual noise used can then be seen in the result textfiles). Receivers that has no indicated individual noise will use the constant background noise from Env. while individual noise will be the given values as above plus power addition of the residual noise in Env. Examples:

10 11

1 3 1.2 uses the constant background noise 0 3 1.2 <50 60 70 70 70 70> uses the given noise + residual noise, 8k an 16k extrapolated

116


12 -1 3 1.2 -1 5 1.2 <50 60 70 70 70 70> uses the given noise + residual noise, 8k an 16k extrapolated 13 -2 3 1.2 <50 60 70 70 70 70 : 60 50> 8k and 16k also given, uses the given noise + residual noise Note: if individual receiver background noise has been used it will apply also in Source addition until Bkg noise change is selected where all receivers will get the selected noise.

OVERRIDE

when using receiver loops it sometimes will place a few receivers inside say pillars and similar. Adding this keyword allows positions created in a loop to afterwards be overridden by a new explicitly given position.

Note: if SCALE is used in all GEO-files as well as in the receiver- and source-files using x(), y(), and z() functions will scale twice since scaling has already been performed. In such a case use e.g. 1+x(12)/s, where s is the scale factor, to prevent the extra scaling. To create receiver coordinates with some repetition or with other inherent logic the recloop(), recloop2() or recwalk() functions can be used. These functions are used instead of an id r_x r_y r_z or id r_x r_y r_z h_x h_y h_z sequence. recloop() syntax (on one line): recloop(id_start,i,i_start,i_stop,i_step,x_expr,y_expr,z_expr)

id_start

i

i_start i_stop i_step x_expr y_expr x_expr

start of receiver ids to be generated by the function, for each new coordinate this number is incremented by 1 loop variable (must be chosen not to clash with a constant name used in any of the expressions for x, y, and z) integer start value for I integer stop value for I integer step for i, can be < 0 for backwards stepping expression for the x-coordinate (optionally including loop variable i ) expression for the y-coordinate (optionally including loop variable i ) expression for the z-coordinate (optionally including loop variable i )

Example: to generate a matrix of say 5 lines of 8 receivers, 5 recloop() can be used instead of 5x8=40 lines of individual receiver definitions, for example: RECEIVERS recloop(00,q,0,7,1,-6,2.0+q*2,1.2) recloop(10,q,0,7,1,-3,2.0+q*2,1.2) recloop(20,q,0,7,1, 0,2.0+q*2,1.2) recloop(30,q,0,7,1, 3,2.0+q*2,1.2) recloop(40,q,0,7,1, 6,2.0+q*2,1.2)

;creates ;creates ;creates ;creates ;creates

00..07 10..17 20..27 30..37 40..47

An advanced use could be to combine a loop with string constants so that the step size is easy to change:

117


LOCAL expr = " -6 , 2.0+q*2 , 1.2 " RECEIVERS recloop(00,q,0,7,1,%expr%) recloop(10,q,0,7,1,%expr%) recloop(20,q,0,7,1,%expr%) recloop(30,q,0,7,1,%expr%) recloop(40,q,0,7,1,%expr%)

recloop2() syntax (on one line): recloop2(id_start,i,i_start,i_stop,i_step,x_expr,y_expr,z_expr, x_expr2,y_expr2,z_expr2)

This syntax is identical to that of recloop() except that three additional expressions are given for the individual head directions, e.g.: RECEIVERS recloop2(00,q,0,7,1,-6,2.0+q*2,1.2,-7,2.0+q*2,1.2) recloop2(10,q,0,7,1,-3,2.0+q*2,1.2,-4,2.0+q*2,1.2) recloop2(20,q,0,7,1, 0,2.0+q*2,1.2,-1,2.0+q*2,1.2) recloop2(30,q,0,7,1, 3,2.0+q*2,1.2, 2,2.0+q*2,1.2) recloop2(40,q,0,7,1, 6,2.0+q*2,1.2, 5,2.0+q*2,1.2)

A loop-like function recwalk(), is made to simplify good receiver placements for real time walkthrough auralization using CATT-Walker, see Post-processing and Walker.PDF. The function creates concentric circles of receivers around a defined point, with selected angle and radius step and confined within a defined rectangle. recwalk() is meant to be used when walkthrough around a source is desired since then the receiver density has to be high, it is also advisable to place four receivers around the source position inside the first circle. Further out in the room, or if walkthrough close to the source is not interesting, a rectangular receiver grid works well and can be created by a couple of recloop() functions. recwalk() syntax (on one line): recwalk(id_start,x0_expr,y0_expr,da_expr,dr_expr,xmin_expr,ymin_expr, xmax_expr,ymax_expr,z_expr)

id_start x0_expr y0_expr da_expr dr_expr xmin_expr ymin_expr xmax_expr ymax_expr z_expr

start of receiver ids x-coordinate of the circle center, typically the source y-coordinate of the circle center, typically the source angle step in degrees radius step lower limit of receiver x coordinates lower limit of receiver y coordinates upper limit of receiver x coordinates upper limit of receiver y coordinates receiver z coordinates

118


Notes for all loops : ·

decimal comma cannot be used in the x-, y- and z- expressions but only decimal point (comma is used to separate parameters).

·

the selected loop variable must not clash with any other used letter in an expression (if e.g. sin() is used, s, i or n cannot be used as a loop variable) .

·

if an error occurs the displayed loop arguments will not look exactly as entered since all blanks have been removed (if lock() was used the blanks between the ids are replaced by a colon (:) like lock(1 2 3) will become LOCK(1:2:3)). This is done to simplify parsing.

·

just like with individually entered receiver coordinates, any SCALE directive in effect will scale the values after they have been created.

Example The example below is the CATTDATA-folder\MODELS\EXAMPLE\REC.LOC. See fig 2.9.

;CATT-Acoustic v8.0 Manual Example Hall LOCAL dx = 1.9 ; column spacing LOCAL x0 = 0 LOCAL x1 = x0 + dx LOCAL x2 = x1 + dx LOCAL x3 = x2 + dx LOCAL x4 = x3 + dx LOCAL x5 = x4 + dx LOCAL dy = 7 ; row spacing LOCAL y0 = 5 LOCAL y1 = y0 + dy LOCAL y2 = y1 + dy LOCAL y3 = y2 + dy LOCAL dz = 0.5 ; height spacing LOCAL z0 = 0.6 LOCAL z1 = z0 + dz LOCAL z2 = z1 + dz LOCAL z3 = z2 + dz RECEIVERS 1 -x4 y0 2 -x3 y0 3 -x2 y0 4 -x1 y0 5 x0 y0 6 x1 y0 7 x2 y0 8 x3 y0 9 x4 y0 10 -x4 y1 11 -x3 y1 12 -x2 y1 13 -x1 y1 14 x0 y1 15 x1 y1 16 x2 y1 17 x3 y1

z0 z0 z0 z0 z0 z0 z0 z0 z0 z1 z1 z1 z1 z1 z1 z1 z1

119


18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

x4 -x3 -x2 -x1 x0 x1 x2 x3 -x3 -x2 -x1 x0 x1 x2 x3

y1 y2 y2 y2 y2 y2 y2 y2 y3 y3 y3 y3 y3 y3 y3

z1 z2 z2 z2 z2 z2 z2 z2 z3 z3 z3 z3 z3 z3 z3

120


2.4 Output-files This section contains a list of all result plot- and text-files created by the Prediction module. It complements the general overview of file-types in Section 1.4. All files except *.TXT are in binary format. Most kinds of echogram graphs display a dual time axis, the upper axis is calculated from the direct sound arrival, the lower from the emission of sound. Note that the lower axis is rounded to ms so if the actual initial delay is e.g. 7.2 ms the lower axis will start at 7 ms.

General PLT… .TXT

Each time PLT-files are created a list of the files are created and loaded into the PLT viewer module. This plot-file list is named according to the combination of actions requested by concatenating short descriptive names: PLT _GEO _MAPD _MAPDR _ISM _FULL so that with e.g. a Geometry view/check plus Full detailed calculation the plot-file list will be named PLT_GEO_FULL.TXT. Plot-file lists can also be loaded via List|Open List-file.

Geometry view/check results Selected in the Geometry view/check dialog:

VIEW4.PLT

Plan, side, and top views plus a parallel 3D projection. Sources and receivers are marked. If audience planes are selected for Audience area mapping they are marked grey and the audience surface-area is displayed. For closed models the estimated volume is also displayed (if a model is said to be closed in Geometry view/check dialog but found to be open, the volume is displayed in round brackets). If only a geometry check is performed, the volume estimate is less accurate than for full calculations and Interactive RT estimate.

SRCINFO.PLT

Similar to VIEW4.PLT but instead of a 3D projection data for each source selected is shown in plot sub-frames. If a source is not omni-directional the source visual is shown. Source aim angles are defined as in fig:

z

-180° ≤ H ≤ 180° -90° ≤ V ≤ 90° H=0°,V=0° is the y-axis H > 0° on the positive x side V > 0° on the negative z side

V x

H

y

aim line

COLORED.PLT

A colored 3D projection of the hall. Sources and receivers are marked. Head directions (if not towards source) and source aim (if not Omni directivity) are marked as a line pointing out from the receiver or source. The reflecting side of surfaces are colored according to the assigned surface colors. If a source is not omni-directional the source visual is shown.

SHADED.PLT

A shaded 3D projection of the hall. Sources and receivers are marked. Head directions (if not towards source) and source aim (if not Omni

121


directivity) are marked as a line pointing out from the receiver or source. The reflecting side of surfaces are given a viewing-angle dependent shading. If a source is not omni-directional the visual is shown.

SHADED.OGL

A shaded 3D projection of the hall for the OpenGL-based CATT 3D-viewer. Similar to COLORED.PLT but with more 3D functions, lighting, a walkthrough option, animated viewpoint transitions, optional -3, -6 and --9 dB directivity contours etc. The source visual is shown. The 3D-viewer allows for sub-division selection with data display and selection of a list of planes, this list of plane ids can be pasted into a plane set edit (Free form edit) dialog such as e.g. Audience planes in the Audience are mapping dialogs.

COINFO.PLT

A wire-frame model with all corner ids displayed (no shading). Note: a hidden option will enable showing also ids not used in any plane.

PLINFO.PLT

A wire-frame model with data on all planes included as plot sub-frames. All planes must have their front sides (the side that can be hit by sound) colored with the assigned surface color and the back be black (or the plane corner entering order must be reversed). Absorption coefficients are shown in black, scattering coefficients (surface and auto-edge) in red, transmission coefficients are shown under the absorption coefficients in green and in the graph by a dotted green line. Some planes have codes (e.g. -o), see DEBUG.TXT below. The name of the GEO-file and the line number where the plane was defined name is given as FILENAME:nnn.

REFLINFO.PLT

A wire-frame model with data on all selected planes included. Each selected plane is a plot sub-frame showing edge reflection contours from each source. Note that the contour is purely geometrical and no consideration to diffraction or diffusion is taken. Contours for not selected planes are shown gray.

SRCINFO3D.PLT Similar to SRCINFO.PLT but a 3D wire-frame model where the -3dB contours are shown for each source for the selected octave-bands. Contours for not selected sources are shown grey. name.CMB

Optionally created source combination file for the Source addition module (automatically created if ADD-files are selected in Early part detailed ISM or Full detailed calculation. name is taken from the current source-file).

DEBUG.TXT

Optional text-file describing the geometry in detail as well as any geometrical errors found, see Section 2.2. Planes listed have codes describing the type of plane: -c (concave plane i.e. with reentrant angles), -o (obscuring plane i.e. it can intercept a ray between two other planes), d (double-sided plane - the other side's plane id follows). Plane subdivisions have codes -c (concave sub-division), -* (has diffusing properties). If a list of planes is selected the plane corners of concave planes are, for debugging purposes, indictated by x (used in plane equation), * (on the convex hull) and - (not on the convex hull). Example:

CATT-Acoustic v8.0a debug : Sample shoebox ---------------------------------------------------------------------DUPLICATE PLANE ID's : no ! DUPLICATE CORNER ID's : no ! DUPLICATE CORNERS : no ! SINGLE-CONNECTED CORNERS : no ! INACCURATE PLANE CORNERS : no ! EDGES CUTTING/TOUCHING : no ! POSSIBLY REVERSED PLANES : no ! ----------------------------------------------------------------------

122


---------------------------------------------------------------------ID x y z Used in planes ---------------------------------------------------------------------1 -5,000 0,000 0,000 1 3 5 2 -5,000 24,000 0,000 1 4 5 … 13 5,000 24,000 8,000 2 4 6 14 5,000 0,000 8,000 2 3 6 Planes used (* = diff) :

1*

2*

3*

4*

5*

6*

Plane ID: 1 Name : floor (S= 240,0m²) Corners : 4 3 2 1 Equation: 0,0000 0,0000 1,0000 0,0000 Abs/Diff:AUDIENCE <40 50 60 70 80 80 : 80 : 80> L <30 40 50 60 70 80 : 90 99> Plane ID: 2 Name : ceiling (S= 240,0m²) Corners : 11 12 13 14 Equation: 0,0000 0,0000 -1,0000 8,0000 Abs/Diff:WOOD <15 13 10 9 8 7 : 6 5> L <30 30 30 30 30 30 : 30 30> …

Audience area mapping results Selected in the Audience area mapping dialogs: The calculation case for mapping calculations is documented in the plot-file title (the yellow bar below the plot) by a string:

N/tr/height/P/A/F Where:

N tr height /P /A /F

number of rays ray truncation-time, ms map height over the selected audience planes, m if direct sounds are added with phase if Adapt rays is in effect some source used a MAPRAYFACTOR (see source-file)

For example: 10000/1300/0.5/P means 10000 rays traced for 1300 ms and a map height of 0.5 m, direct sounds added with phase.

SPL_oct.PLT

Sound pressure level coverage maps for the time-intervals indicated.

SPLDIR_oct.PLT Direct sound pressure level coverage maps for the time-intervals indicated. SPL_oct_n_3D.PLT 3D SPL coverage map for time-interval n. SPLDIR_oct_n_3D.PLT 3D Direct sound SPL coverage map for time-interval n. LF_oct.PLT

LF coverage maps for the time-intervals indicated.

123


LF_oct_n_3D.PLT 3D LF coverage map for time-interval n.

LFDIR_oct.PLT

Direct LF coverage maps for the time-intervals indicated.

LFDIR_oct_n_3D.PLT Direct 3D LF coverage map for time-interval n. RASTI.PLT

RASTI coverage map both with and without background noise.

RASTI_with_noise_3D.PLT 3D version. RASTI_without_noise_3D.PLT 3D version. STI_name.PLT

STI coverage map both with and without background noise. STI type can be selected in Preferences, depending on type chosen the name part of the file changes: orig, mdfd, IECm, IECf or user.

STI_name_with_noise_3D.PLT 3D version. STI_name_without_noise_3D.PLT 3D version. PARn_oct.PLT

Parameter maps (n = 1,2,3) as selected.

param_oct_3D.PLT 3D for param (e.g. D50).

PARnDIR_oct.PLT Parameter maps direct sound only (n = 1,2,3) as selected. paramDIR_oct_3D.PLT 3D for param (e.g. SPL).

NOISEMAP_oct.PLT With noise-mapping the actual noise SPL map created by the defined noise sources. NOISEMAP_oct_3D.PLT 3D version. For A-weighted results the oct part of the syntax is “asum” On each color scale min and max lines are drawn. Audience area mapping statistics

When mapping is requested for any of the measures D-50/C-50, C-80, LF, SPL, RT', Ts, RASTI or STI, map statistics is automatically written to a text-file MAPSTAT.TXT in the output folder plus a version made for spreadsheet import (semicolon delimited) MAPSTATx.TXT (and shows up in the Prediction:Latest results dialog as do all created text-files). This file first lists information about all sources used and then the percentage of map points having predicted values in different intervals together with simple bar graph for a quick overview (one * = 1% rounded). An example for LF:

124


------------------------LF 2 kHz [%] ------------------------< 5.0 : 0.83 % * 5.0 - 10.0 : 2.50 % *** 10.0 - 15.0 : 8.75 % ********* 15.0 - 20.0 : 13.75 % ************** 20.0 - 25.0 : 12.92 % ************* 25.0 - 30.0 : 18.33 % ****************** 30.0 - 35.0 : 31.25 % ******************************* 35.0 - 40.0 : 11.25 % *********** 40.0 - 45.0 : 0.42 % 45.0 - 50.0 : 0.00 % > 50.0 : 0.00 % -------------------------

The choice was made to make a simple text-file rather than a plot-file graph since some users anyway may want to cut/paste this information into a report table. The internal implementation allows for uneven sized intervals (e.g. for RASTI, see below) but so far this is not reflected and intervals are simply set to reasonably well correspond to the difference limen of each measure: Parameter Min Step Max Unit Comment D-50 10 10 90 % i.e. ranges: <10,10-20,..,80-90,>90 C-50 -10 1 10 dB C-80 -10 1 10 dB LF 5 5 50 % See example above SPL max-20 1 max dB top 20 dBs are divided in 1 dB steps G max-20 1 max dB see SPL RT' min 0.1 max s Ts min 10 max ms RASTI 30 15 75 % i.e.: <30,30-45,45-60,60-75,>75 *) STI 30 15 75 % see RASTI *) Corresponds to the BAD, POOR, FAIR, GOOD, and EXCELLENT ranges. Note: if a map is made in such a way that a large fraction of the map points are partly out of walls or are not hit at all because of blocking it has to be taken into account by comparing with the map plot-file (mostly happens with a large map step). The statistics evaluates how many of the map points that have been hit at all fall into the categories given. Note: audience area mapping results are always for the sum of all sources selected including any source delays. If noise-mapping is used a special MAPSTAT_NOISEMAP.TXT is created. If the statistics of the noise-map SPL is quite even the average can be used as a Total average Background noise in General settings and the noise-mapping and the noise sources be switched off.

Early part detailed ISM results Selected in the Early part detailed ISM dialog: The calculation case for early part ISM calculations is documented in the plot-file title (the yellow bar below the plot) by a string:

S/D/te

125


Where: S D te

Specular reflection order Diffuse reflection order (or "-" if diffusion is off) Truncation-time, ms

For example: 4/1/150 means 4 orders of specular reflection, 1 order of diffuse reflection and early part truncated at 150 ms.

I_ss_rr_oct.PLT Echogram plot-files for each octave. H_ss_rr_oct.PLT Echogram and interactive specular reflection trace (first order diffuse reflections are colored red). In the lower right graph plane ids that are encountered by each reflection is displayed (S = source and R = nd receiver so that S-12-10-R means a 2 order reflection in planes 12 and 10). The lower left shows source exit angles as the ray (line length affected by directivity) projected in two planes. The upper right shows the incidence angles to the receiver displayed in a similar manner (line length affected by reflection strength). Note: the 3D view of the model can be rotated unless the 2D version of this plot is selected in Preferences. I_ss_rr.TXT

Hidden option (see Section 1.3): A text-file with all reflections are created. The file is intended for those that may want to do a postprocessing of their own for some special purpose, 8k and 16k values are included according to Preferences. The file has the following structure (decimal point or comma is used depending on Windows Control-panel Regional settings):

#HEADER (header data similar to PARAM_ss.TXT, skip to #DATAFIELDS if not required) #DATAFIELDS TYPE DELAY ORDER AZIMUTH ELEVATION SPL125 SPL250 SPL500 SPL1K SPL2K SPL4K SPL8K SPL16K IS_x IS_y IS_z (on one line)

(these fields indicate which fields are in the #DATA section below, and in what order: TYPE

is d = direct sound, s = specular, * = diffuse (not sorted in time);

DELAY

is delay relative direct sound arrival in ms;

ORDER

is reflection order (diffuse ones always have 1);

AZIMUTH

is the standard azimuth relative the listener head (affected by headdirection);

ELEVATION is the standard elevation relative the listener head (affected by headdirection); x-axis is left to right, y-axis is in the nose direction, and z-axis is up, frontal incidence will thus be at 90,90 degrees. Look at the history plotfile in parallel if in doubt about the angle values; SPL...

is the SPL for the corresponding octaves);

IS_x,y,z

image source location in in m.

#DATA

126


d 0.00 0 90.00 90.00 s 0.96 1 90.00 101.81 s 5.81 1 90.00 89.20 (rest of the data)

69.1 67.3 66.7

69.1 67.3 66.7

69.1 67.3 66.7

69.1 67.3 66.7

69.1 67.3 66.7

69.1 ... 67.3 ... 66.7 ...

#END

S_ss_rr_oct.PLT Image source space plot-files where the circle area is proportional to the image source power. The lower right graph displays the estimated late part reflection growth together with actual early part growth (using classical theory for the late part requiring a closed shape). In each view are shown only those image sources that are contained in a slice of space described in the other views. For example, in the x-y view only the sources between the horizontal dotted lines in the y-z view are shown so that they represent what is heard in the horizontal plane. E_ss_rr.ECH

Echogram files for the post-processing module (only for special cases).

E_ss_rr.ADD

Echogram files for the Source addition module (only for special cases).

name.CMB

Source combination file for the Source addition module automatically created if ADD-files are selected. name is taken from the current source-file.

Full detailed calculation results Selected in the Full detailed calculation dialog The calculation case for full detailed calculation is documented in the plot-file title (the yellow bar below the plot) by a string:

N/tr Where: N tr

Number of rays Ray truncation-time, ms

For example: 2000/1500 means 2000 rays traced for 1500 ms.

E_ss_rr_oct.PLT

Echogram plot-files for each octave with extensive information:

upper left: ·

full echogram (black curve) with backward integrated decay (red curve), reverberation time regression lines and coefficients;

upper right: ·

early part discrete reflections: direct sound (blue bar with a ring), first and second order specular reflections (blue bars), first order diffuse reflections (red bars), all other reflections (black bars);

·

early part smoothed echogram (lower red curve, filter can be selected);

·

early part backward (upper red curve) and forward integrated (upper black curve), the crossing of the forward and backward integrated curve will be an estimate of the “Rise time” as defined by W.J. Jordan [Cremer, p. 433];

lower right: ·

2

early part cos -weighted X (front-back, blue curve), Y (left-right, black curve) and Z (up-

127


down, red curve) smoothed echograms; lower left: ·

a small hall plan with source- and receiver locations, a scale and source data;

·

all estimated major parameters.

As a visual aid to understanding how noise affects speech intelligibility, if any of the measures STI and RASTI (or U50, AI, PI or SII as Hidden options) are chosen, the background noise is plotted as a heavy dashed line in echograms.

E_ss_rr_oct.PLT and similar echogram plots show a dual time axis. The upper axis gives the time relative arrival of direct sound while the lower gives the time relative sound emission. E_ss_rr_INT.PLT Backwards integrated decays (“Schröder plot”) for all octaves marking T-15, T-30 and their regression lines. R_ss_rr_oct.PLT Sound rose plots indicating sound incidence projected onto three planes and in six time-intervals. Within each time range the incident ray vectors are projected (the length of the vector is its intensity value) onto the horizontal plane, the vertical plane and the ear-to-ear vertical plane where they are accumulated in 5° wide sectors and a 10 dB/division radial scale is used. Each rose shows only the spatial incidence and not the absolute level since the length of the lines are normalized by the total incident sound energy in each the time-interval. At each rose a % value is given calculated as the ratio of the energy projected to each plane to the total energy in the time-range. V_ss_rr_oct.PLT Vectorgram plots, an echogram where each reflection is drawn according to the incidence. Imagine standing at the origin looking along the time axis. Direct sound and first order specular reflections are drawn with thick lines. E_ss_rr_EKgrad.PLT Echo-disturbance plots according to [Dietsch&Kraak-86]. The created graphs show estimated 10% and 50% echo-disturbance for speech (red) and music (blue) overlaid on the echogram for each octave (gray). The right side scale indicates EKgrad and goes form 0 to 2, for details see Theory. Note: EKgrad values sometimes exceed 2.0 but it is a benefit to keep a fixed scale so overshooting values are plotted just above the graph frame at 2.1 (a value of 2.0 is already very high). RT_ss.PLT

Reverberation time and mean absorption overview based on an average over all receivers used. Also indicates the number of rays lost and absorbed per octave-band and other statistical data. If the two mean absorption coefficient curves (AbsC and AbsCg) differ it is likely that overlapping planes have been used or that surfaces that can never be reached by sound have been modeled.

PATH_ss.PLT

Histograms over free reflection path lengths and the resulting mean free path.

ABS_ss.PLT

Histograms over absorption coefficients encountered by the rays and the resulting mean absorption coefficient.

HITS_ss.PLT

Interactive plane hit histogram. Each plot frame marks each plane's share of all ray hits. Above the graph is displayed each plane's area efficiency (for all eight octaves), its share of the total absorption area and its classical absorption area. For area efficiency, a value of 100%

128


indicates that a surface is hit according to classical Sabine theory (areaproportional), < 100% means hit less often and > 100% means hit more often.

PARAMn_ss.PLT

Parameter overview plot-files (n = 1,2,3). The data for each octave is a plot frame. If a value is too high to display with three digits ">>>" is shown and if a value is too low "<<<" is shown. If a value could not be estimated "***" is shown and if it is not applicable "---" is shown.

TRENDn_ss.PLT

Parameter trend plot-files (n = 1,2,3). A plot frame line can be moved marking the receivers. Receivers not used are indicated by a small ring and values that could not be estimated are indicated by a small square.

NTREND_ss.PLT

Similar to TRENDn_ss.PLT but with normalized parameters, see Theory.

RASTI_ss.PLT

Trend of RASTI values, 500Hz/2kHz average D-50 (Deutlichkeit), and 500 Hz/2 kHz SPL (RASTI is built from these two octaves). Receivers not used are indicated by a small ring. Background noise levels are indicated. Note: See Preferences option for STI weights.

STI _type_ss.PLT Trend of STI and octave band TI (Transmission Index) values. Receivers not used are indicated by a small ring. Background noise levels are indicated. Depending on the STI Preferences selection type will be orig, mdfd, user, IECm or IECf. SPECTRA_ss.PLT

A plot of the SPL octave-band spectra for direct sound, first 50 ms and total as an aid in overall system eq and coverage. Spectra for all receivers are shown in light grey and each receivers data is as plot frame (a 2DA plot-file) also indicating a line from the source to the receiver (dashed if there is no direct sound). A reference or target spectrum can be given for reference (plotted as rings with a 1 dB radius).

LEAKS_ss.PLT

If any ray leaks were encountered during the tracing of rays, this file is automatically created showing where the leaks occurred. At the top of the largest leak, the number of rays that escaped through that part of the model is indicated. Note: the direction of a ray is not recorded but only the location where it escaped. Causes to leaks are warped planes, actual holes and a finite numerical accuracy. In addition is displayed a graph showing the fraction of lost rays over time.

Walker.CAG

This file is always created by the full version (CAG = CATT-Acoustic Geometry) if (Save data for) Post-processing was selected. It is used by CATT-Walker to display the room geometry during a walkthrough, it is also needed for post-processing to create a CWI-file for CATT-Walker.

E_ss_rr.TXT

Single receiver result text-file. If Include parameters normalized to expected values has been selected the normalized values are shown alongside with the absolute values. Example:

CATT-Acoustic v8.0g : Acoustic parameters Copyright © CATT 1988-2007 ============================================================================================ Original file name: C:\CATTDOCS\V8\EXAMPLE\OUT\E_A0_01.TXT Original save time: 2007-06-22 15:23:51 -------------------------------------------------------------------------------------------Project : v8.0 Manual Example Hall

129

2. PREDICTION MODULE GEOMETRICAL INFORMATION Src id and loc [m] Rec id and loc [m] VARIABLE SOURCE DATA Directivity Type (library) Dir. Index (DI) [dB] Level On axis 1m [dB] Total power [dB] Auxiliary delay [ms]

VARIABLE AIR DATA Temperature [°C] Relative humidity [%] Density [kg/m³] Sound speed [m/s] Impedance [Ns/m³] Diss. coeff. [0.001/m]

: :

: :

A0 01

3,000 -7,600

OMNI.SD0 0,0 0,0

: : :

70,0 80,9 0,0

: : : : : :

20 50 1,2 343 412 0,10

-2,000 5,000

1,700 0,600

0,0

0,0

0,0

0,0

0,0

0,0

73,0 83,9

76,0 86,9

79,0 89,9

82,0 92,9

85,0 95,9

85,0 95,9

85,0 95,9

0,30

0,63

1,07

2,28

6,83 24,23 83,90

(estimated)

CALCULATION RESULTS Head direction [m] : 01 head turned towards stage GLOBAL RESULTS FROM TRACING – RTC-II -------------------------------------------------------------------------------------------Trunc[s] Rays[-] Lost[-] Absorbed[-] Angle[°] ............................................................................................ 1,50 10088 0 0 2,0 -------------------------------------------------------------------------------------------125 250 500 1k 2k 4k 8k 16k .......... ..... ..... ..... ..... ..... ..... ..... ..... ................................. MFP [m] 10,45 10,45 10,43 10,44 10,46 10,47 10,47 10,49 Mean Free Path Diffs[%] 15,11 18,71 22,66 27,86 31,76 32,42 33,19 33,91 Average scattering coefficient ............................................................................................ T-15 [s] 1,84 1,82 1,68 1,70 1,59 1,35 0,84 0,34 (LS-fit -5 to -20 dB) T-30 [s] 1,92 1,91 1,79 1,81 1,73 1,47 0,85 0,32 (LS-fit -5 to -35 dB) EyrT [s] 1,61 1,54 1,35 1,29 1,32 1,21 0,77 0,36 (MFP, AbsC) EyrTg[s] 1,61 1,53 1,33 1,27 1,30 1,19 0,76 0,35 (MFP, AbsCg) SabT [s] 1,79 1,71 1,52 1,45 1,46 1,30 0,81 0,37 (Vact, Sact, AbsCg) AbsC [%] 22,9 23,7 26,3 27,0 25,6 24,2 25,4 26,3 (based on tracing) AbsCg[%] 23,0 23,9 26,6 27,4 26,0 24,7 26,0 27,0 (area-proportional) Back[dB] 45,0 38,0 32,0 28,0 25,0 23,0 21,0 19,0 (non-individual bkg noise level) Resi[dB] 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 (indiv. resid. bkg noise level) -----------------------------------------------------------------------------------------AVERAGE/ACTUAL VOLUME/SURFACE INFORMATION -------------------------------------------------------------------------------------------V[m³] Lx[m] Ly[m] Lz[m] S[m2] Sact[m²] Vact[m³] ............................................................................................ 7892 21,0 31,7 11,9 3019 3019 7747 V Lx,y,z S Sact Vact

is volume estimated from MFP and S is an average size of the model is surface area estimated by the rays is total surface area is actual estimated volume (requires a correct model)

130

2. PREDICTION MODULE --------------------------------------------------------------------------------------------

Initial delay

[ms] :

37,14

--------------------------------------------------------------------------------------------

RASTI [%]

56,3 (FAIR)

56,3 (FAIR)

with background noise

STI orig [%] 57,6 (FAIR) 57,2 (FAIR) with background noise -------------------------------------------------------------------------------------------125 250 500 1k 2k 4k 8k 16k ........... ...... ...... ...... ...... ...... ...... ...... ...... TI [%] 54,6 52,7 56,2 56,7 56,0 59,2 67,5 TI(n)[%] 52,0 52,4 56,2 56,6 56,0 59,2 67,5 Weights 130 140 110 120 190 170 140 x 0.001 (orig.) -------------------------------------------------------------------------------------------Parameter 125 250 500 1k 2k 4k 8k 16k sum ........... ...... ...... ...... ...... ...... ...... ...... ...... ...... Ts [ms] 92,2 94,3 82,3 82,1 82,8 73,4 47,0 23,2 69,8 D-50 [%] 52,6 46,0 52,1 51,5 52,6 52,4 68,1 88,0 56,9 C-80 [dB] 3,1 2,5 3,5 3,9 3,1 3,8 7,1 14,3 4,4 LFC [%] 47,5 48,8 48,4 45,1 46,0 49,4 46,5 33,1 46,6 LF [%] 35,6 36,5 36,3 32,8 34,6 36,6 34,9 24,9 34,7 G [dB] 7,4 7,4 6,7 6,4 6,2 6,7 4,4 1,3 5,6 SPL [dB] 57,4 60,4 62,7 65,4 68,1 71,4 68,3 62,7 75,6 75,6 (A) (previous line) EDT [s] 1,62 1,43 1,21 1,23 1,28 0,98 0,62 0,34 --T-15 [s] 1,84 1,82 1,68 1,70 1,59 1,35 0,84 0,34 --T-30 [s] 1,92 1,91 1,79 1,81 1,73 1,47 0,85 0,32 ---

--- not calculated *** impossible to calculate with accuracy PARAM_ss.TXT

Multiple receiver result text-file. Similar to E_ss_rr.TXT but for all receivers. Always created.

PARAM_ssX.TXT Multiple receiver result text-file to import into spread-sheet programs e.g. MS Excel. Similar to PARAM_ss.TXT but tabular data is semi-colon delimited and text data is quoted. Always created. E_ss_rr.ECH

Echogram files for the post-processing. The basic contents can be checked by a Post-processing utility.

E_ss_rr.ADD

Echogram files for the Source addition module. The basic contents can be checked by a Source addition utility.

name.CMB

Source combination file for the Source addition module automatically created if ADD-files are selected. name is taken from the current sourcefile.

131


2.5 Theory Prediction methods CATT-Acoustic offers three independent prediction, or acoustic simulation, methods detailed in this section: Audience area mapping

utilizes standard ray-tracing with a spherical receiver. Ray-tracing is a robust method for prediction of numerical measures but the echograms are difficult to use for auralization since the reflection density growth over time is unnatural.


utilizes the Image Source Model with added first-order diffuse reflection. This method is meant for qualitative reflection path analysis and does not estimate any room acoustic parameters. The reason is that diffuse reflection is extremely difficult to include in the ISM in a general manner. For special cases that require high early part detail but no reverberation, postprocessing files can be generated.


utilizes Randomized Tail-corrected Cone-tracing (RTC) that combines features of both specular cone-tracing, standard raytracing and the ISM. The RTC is a general and robust numerical prediction method and can as well create echograms that can be used for auralization. Since this method (as all other prediction methods) has drawbacks, the direct sound, first order diffuse and specular reflections and second order specular reflections are handled deterministically by the ISM.

The predictions method are based on geometrical acoustics and are partly documented in a Ph.D. thesis [Dalenbäck-95b], [Chalmers].

Speed of sound and air absorption The speed of sound and air absorption is calculated according to the following expressions:

ϑ ) 273 mb = from ISO 9613-1 (6.2 Formulae) based on ϑ, fb, and Φ

c0 = 331.4 1 +

(2.5-1) (2.5-2)

where ϑ mb fb Φ

air temperature, °C -1 air dissipation coefficient, m (see [Cremer] p213, mb = 2m) octave-band center frequency, Hz relative humidity, %

132


Audience area mapping Audience area mapping estimates early sound pressure level, lateral energy fraction, and parameters (D-50 etc.), or direct sound only over a grid covering selected audience-planes. A separate ray-tracing with fixed-size spherical receivers, where also source directivity and source delay is taken into account, is performed. "fixed-size receiver" means that, unlike conetracing, the receiver radius does not change depending on how long a receipted ray has traveled. Diffuse reflection is handled by randomizing the direction of rays reflected off diffusing surfaces and the likelihood of diffuse reflection is determined by the scattering coefficient, s, of each surface. The map is divided into a square grid where the grid step can be selected in the Audience area mapping dialog. The receiver diameter is set to slightly larger than the map step and the receiver center is normally placed at half a map step above the mapping planes so that the sphere does not penetrate the audience plane. Some squares may fall partly outside the room (close to wall positions) causing a loss of energy received at those positions. To arrive at correct pressure levels the number of rays used should ideally match the map step chosen. To avoid this strict dependence, a different approach has been used. The number of rays needed to give a possibility of one ray per square at the longest chosen time is first calculated. The power of each ray is then adjusted dependent on the actual number of rays used so that roughly the same levels are obtained even if the number of rays are "too few". A systematic error is thus converted to a statistical error. If audience surfaces overlap in z, only the surface with highest z is mapped in 2D map views (i.e. if both balcony and stalls are included the stalls under the balcony will not be mapped but has to be predicted separately). For static 3D map views the map points closest to the viewpoint are drawn last thus covering points behind. If special separate 3D maps are chosen all audience planes will be seen as the model is rotated. The direct sound is handled separately connecting the source directly to each receiver center taking into account any obscuring surfaces and is not effected by the number of rays chosen.

Early part detailed ISM The Image Source Model (ISM) can determine all specular reflections between a source point and a receiver point. The ISM works by calculating first order image sources (IS) of the main source in all reflecting planes. From each of these first order IS, second order sources are created by calculating new IS in all reflecting planes (except the plane that created the first order IS since it would yield the previous order IS). This procedure is repeated until a requested max order of reflection or max arrival-time is exceeded. Once all valid IS are found, the corresponding reflection arrival times can be found simply by calculating the length of the three-dimensional vector from each IS to the receiver. The level of each reflection is calculated from the output power and directivity of the source, spherical spreading (inverse square law) and is adjusted by absorption and diffusion (first order) properties of the planes that created the IS and also by air absorption. On each diffusing surface a number of square patches are placed. For first order reflections these patches act as elementary sources radiating according to Lambert's law when excited [Kuttruff]. Diffusing surfaces are automatically patched. The Early part detailed ISM dialog selection Diffuse refl. order determines the maximum order that will give diffuse reflection. All specular reflections up to and including the diffusion order will be diminished by (1 - s). For higher order images sources the scattering coefficient, s, is neglected (or the energy would be lost). The diffuse reflection order should typically be set to 1.

133

2. PREDICTION MODULE How transparency is taken into account

·

direct sound passing through a maximum of one semi-transparent surface is attenuated by (1-α) τ. For higher orders, transmission is treated as an additional absorption only.

·

1 order diffuse reflections are not transmitted (the transmitted sound would be extremely weak) - the transmitted part is assumed absorbed.

st

Full detailed calculation CATT-Acoustic v7 and v8 utilize a unique prediction method that is based on many years of experience from using the ISM, ray-tracing and combinations thereof, in previous software versions as well as on a Ph.D. thesis at Chalmers University [Dalenbäck-95b], [Chalmers].

The design goals for the new method, called Randomized Tail-corrected Cone-tracing (RTC for short, ≥ v7.2 uses RTC-II), were to create: ·

a general method that requires as few assumptions as possible e.g. regarding the statistical properties of the room under consideration:

RTC only assumes that reflection density growth is quadratic and that can be shown to hold generally for closed rooms with no coupling effects. For open rooms no use is made of the reflection growth. ·

a method that uses the same algorithm for both the early part and the late part of the echogram:

RTC-II employs randomized cone-tracing (cone-tracing that handles diffuse reflection in the same manner as ray-tracing does) for the full response but handles the direct sound, first order specular and diffuse reflection, and second order specular reflection deterministically. What is labeled the "early part" in RTC is up to the time where details of the echogram are considered to be valid and no extrapolation of the growth is necessary (due to the divergence of rays cone- and ray-tracing do not detect all late reflections from small surfaces). The more cones that are used the longer the "early part" becomes (typically 300-500 ms in concert halls, i.e. longer than in most other combined methods). The early part is limited to 1 s. ·

a method that can handle frequency dependent diffuse reflection for the full echogram and not only for the late or the early part (this rules out pure ISM, specular cone- or pyramid- tracing, the hybrid method etc.):

RTC handles diffuse reflection for order > 1 the same way as ray-tracing does i.e. for each reflection a random number [0,1[ is generated and if the number is less than the scattering coefficient, s, of the surface the ray direction is randomized according to Lambert's law (ideal diffuse reflection [Kuttruff]) otherwise the reflection is specular. As a necessary consequence of the frequency dependent diffuse reflections, an independent cone-tracing is performed for each octave-band, so when 10000 rays are selected, 8x10000 rays are actually traced. RTC handles diffuse reflection for order = 1 by creating a number of elementary sources on each diffusing surface (the density is governed by s(1-α) giving hard diffusing surfaces the highest density). The radiated power is according to Lambert's law and is proportional to s(1-α) where s is the frequency dependent scattering coefficient of the surface and α the absorption coefficient. Similarly, any first-order and second-order specular reflections (handled by the ISM) are diminished by (1- s)(1- α). This procedure will generate a very great number of weak first-order diffuse reflections. ·

a method that requires few settings (methods with different algorithms for the early and the late part often have settings about the early-late transition that may be very critical and difficult to select):

RTC just requires the selection of the number of rays/cones to use and how long to trace

134


them and even offers an automatic choice. ·

a method that can handle open rooms:

RTC detects the case of an open room and does then not use extrapolation (since the reflection growth is unknown in an open room). Good calculations for open rooms require more rays than for the average closed room but no fundamental method limitations exist. Calculations for open rooms also goes faster since all rays eventually escape. For very big open rooms, such as stadiums, the ISM can be used as an alternative for auralization. ·

a method that can handle rooms with one or two dimensions very different from the third (this rules out many methods that rely on e.g. classical statistical tails):

RTC requires the use of more rays for rooms with odd shapes than for the average closed room but no fundamental method limitations exist (it is quite natural that unusual cases require more rays). However, extremely disproportionate rooms (say with dimension ratios of 1:1:10 and worse) will be less well predicted. Since rooms like these are also less usual there is little feedback from measurements. ·

a method that can resolve under-balcony positions and similar:

RTC requires the use of more rays (i.e. narrower cones) for good prediction of difficult positions such as under balconies but no fundamental method limitations exist (it is quite natural that under balcony positions require more rays because the shape is narrow). However, extremely convoluted rooms where the direct sound is not present, and where the first sound may not reach a receiver until after one or even more reflections, is unlikely to be predicted well since diffraction effects will be considerable. Since rooms like these are also less usual there is little feedback from measurements. ·

a method that is robust and gives reasonable results also with fairly few rays:

Since cone-tracing has an initially small receiver (or cone face) the likelihood for detection of a randomized ray is less than for ray-tracing (using a larger fixed-size receiver) giving a higher statistical variation from run to run. This is compensated for in the RTC by handling the very important first-order specular and diffuse reflections, as well as second order specular reflections, deterministically. Even with as few as 1000 rays, estimated parameters should typically not vary more between runs than the difference limen of each parameter unless the statistical properties are very special (although minor details in the early echogram for reflection order > 1 will differ from run to run). Since a stochastic process is involved it may be necessary to perform test runs for each new hall to determine the number of rays needed to give sufficiently repeatable results (this is no difference from other methods using a stochastic process). ·

a method that can create echograms possible to use for auralization (this rules out direct use of standard ray-tracing since the reflection density is not naturally increasing with time, and specular cone-tracing since it creates unnatural late echoes when no diffusion is applied):

RTC creates discrete reflections as were they created by the ISM and the reflection growth goes as time squared up to the end of the "early part" where reflections are beginning to become undetected. For numerical prediction, energy in the late part (the "tail") is corrected using reflection density growth extrapolation but details are not valid and would give an unnatural-sounding response. For post-processing only, the late part is re-created by statistical generation of reflections (with the correct reflection density growth) using a simplified model of the room taking into account the room shape. The late part ray incidence distribution from the prediction is used for post-processing so that also the late part of the synthesized impulse response has some spatial properties from the actual model. Since the early part often is as long as 300-500 ms the details of the reverberation tail is not so important. If a longer early part is required more rays/cones have to be used which is only natural. This procedure means that late flutter echoes visible in predicted echograms will not be audible in auralization unless they are in the early part. A truthful auralization of the late part would require an extreme amount of rays so at the moment this

135


is the only practical way of having both a natural-sounding late decay and a reasonably fast calculation. Hopefully, cases with late flutter echoes would not need to be auralized since it is so obvious from the echogram graphs that a good room acoustics has not been achieved. For flutter echo demonstration purposes a high order ISM can be used but then only to illustrate the phenomenon. Algorithm option for the late part.

The option differs only for the late part of the echogram where randomized ray-tracing is used instead of randomized tail-corrected cone-tracing (RTC). The benefit is that no tail correction is necessary, and hence no assumptions are required about reflection density growth, and coupled rooms or other unusual shapes can be better predicted. The penalty is a higher random error in the late part and because of that more rays have to be used. However, the new algorithm variant is slightly faster so the total calculation time may be about the same as for the RTC (except for receivers close to a surface, see below). Typical cases where the new option should be chosen are: ·

coupled rooms (rooms connected via open doors, large stage houses, reverb chambers)

· rooms where semi-transparent surfaces are used (see below), option forced on if any semi-transparent surfaces are used ·

unusual shapes (e.g. L-shapes, still requiring lots of rays to resolve unusual shapes and a careful consideration regarding choice of scattering coefficients)

Ray-tracing for the late part uses a receiver sphere (as in audience area mapping) and typically in ray-tracing a fixed-sized receiver sphere is used. However, to create a somewhat smoother first part of the late decay the receiver sphere is allowed to grow up to when it touches the closest surface where the size is locked. This means that if one receiver is very close to a surface, the early part limit may instead be set according to when the sphere cannot grow any longer and be shorter than if the normal RTC is used. In such cases more rays need to be used to get a longer early part. Since this algorithm is designed to handle more complex cases it is only natural that a larger number of rays is required. Do not expect identical results from this algorithm option, reserve it for cases where you suspect unusual behavior such as coupled volumes (say a large reverberant empty stage-house coupled to a hall). How transparency is taken into account

·

direct sound passing through a maximum of one semi-transparent surface is attenuated by (1-α) τ. Higher order transmission is handled by a random process, see below.

·

1 order diffuse reflections are not transmitted (the transmitted sound would be extremely weak) - the transmitted part is assumed absorbed.

·

for higher order transmission sound reflected in a semi-transparent surface is attenuated by (1-α) and the transmission coefficient determines the likelihood for transmission

st

Automatic edge diffusion Reflections falling within a quarter of a wavelength from an edge can optionally be considered not to give a valid specular reflection and the energy can instead be transferred to diffuse. This is controlled globally in the General settings dialog and locally in GEO-files by appending an asterisk "*" to the surface property name in a plane or sub-plane definition for those surfaces that are to give edge diffusion. Edge diffusion is typically applied to reflectors, windows, cupboards, tables etc. that often are flat and smooth and only give significant diffusion if their size is small in relation to the wavelength.

136


Example:

[ 10 reflector / 1 2 12 11 / wood* ] The edge is assigned a scattering coefficient: sedge = 0.5

(50%).

The frequency dependence is implied by the frequency dependent edge width rather than the coefficient itself. For first order reflections edge diffusion is handled as an extra diffuse reflection from the edges so that specular reflections are multiplied by (1 - sedge) and diffuse reflections are multiplied by sedge. For orders > 1 in the RTC, as well as in ray-tracing (audience area mapping), the effective scattering coefficient will be seffective = ssurface + sedge Sedge/Ssurface

Low freq.

Where

Edge

ssurface the normal surface scattering coefficient (if used) Sedge the area of an edge quarter of a wavelength wide Ssurface the complete plane or sub-plane area.

High freq.

If seffective exceeds 100% it is truncated to 100%. This treatment of edge diffusion as an added surface diffusion for orders > 1, can be justified since diffuse reflection then is handled by randomized ray directions and exact impact points are no longer significant. For (too) small surfaces at low frequencies it might result in the whole surface being treated as an edge and they will function as diffusers. Actual edge scattering coefficients (i.e. sedge Sedge/Ssurface) are displayed in the PLINFO.PLT plot-file below the surface scattering coefficients.

Reverberant decay and reverberation time estimation The reverberant decay is estimated from the RTC and by classical methods for reference. There are five measures of the reverberation time calculated: SabT, EyrT, EyrTg, T-15, T-30. The first is a classical Sabine RT, the second two are based on the Eyring formula [Cremer] while the last two are derived from the decays estimated by the RTC for each receiver. SabT is only valid for closed rooms and when surfaces are not overlapping each other (if they are, the total area calculation will be wrong and the covered absorption will be included). Also the volume calculation may be in error, see Fig. 2.6 – 2.9. EyrT is based on the mean free path calculated from all ray segments (i.e. not from 4V/S) and the mean absorption coefficient from the arithmetic mean of all absorption values encountered by the rays, AbsC. EyrTg is based on the mean free path as for EyrT but instead the classical area-weighted mean absorption coefficient is used, AbsCg (g stands for "geometric"). Just like SabT, EyrTg requires non-overlapping surfaces etc. T-15 and T-30 are derived from straight-line least-square fits to receiver decay curves at the intervals -5 to -20 dB and -5 to -35 dB respectively (the regression lines are shown in the

137


decay plot-files E_ss_rr_INT.PLT and in E_ss_rr_oct.PLT). T-30 is considered to be the best estimate of the reverberation time and is used for re-creating the late part in the postprocessing module (if T-30 cannot be evaluated due to a too short echogram T-15 is used instead). The global T15 and T-30 values, as shown in RT_ss.PLT and PARAM_ss.TXT, are calculated as an average for all receiver positions. In addition, Audience area mapping offers a special RT measure called RT’ that is derived from Ts assuming an exponential decay (RT' = 13.8 Ts ; calculated from Eq. 2.5-6 and 2.5-7) and relates well to EDT, see [Cremer]. To estimate a true T-30 for mapping would require an echogram to be recorded at each map point which in turn would mean many MB of data. RT' is often a good estimate of T-30 for positions far from the source. The sometimes large differences between SabT/EyrT and T-30 values are typically explained by a lack of diffuse reflection especially in simply-shaped halls with uneven absorption distribution (a low amount of "mixing" by the shape itself). A good example of this is a shoebox-shaped hall with an absorbing floor where, without diffusely reflecting walls, the reverberation time estimated by Sabine/Eyring and the actual one estimated by the RTC are very different (the ray-decay estimates a much longer RT). This phenomena has been thoroughly investigated in [Kuttruff-88]. The ray-tracing procedure used by Kuttruff is different from the one used in CATT-Acoustic but very similar results are obtained. Also in [Hodgson91] the need for including diffuse reflection in the calculations are emphasized. This behavior can be observed for the simple shoebox example in Section 1.2. Default diffusion (General settings) can be used to assign basic diffusion to surfaces that have not been assigned specific diffusion in the geometry-files. Note: this phenomenon is also present in real rooms, e.g. rectangular sports halls or factories where all surfaces but the ceiling have little absorption. For more details see Calculation recommendations (end of Section 2-5). Another reason why T-30 may be different from SabT/EyrT is if the ray truncation time has been set too short. It should always be set to the longest estimated RT or T-30 will be underestimated. In [ISO] it is said that at least 3/4 of the actual RT must be measured (i.e. down to -45 dB) for T-30 evaluation and it is beneficial to view the prediction as a virtual measurement.

Room acoustic parameters Explanation of how the various room acoustic parameters (measures) are calculated.

Mean free path The mean free path is calculated from all complete reflection paths with reflection order > 1.

Mean absorption coefficient Mean absorption coefficient, AbsC, is calculated from the same reflection paths as the mean free path. AbsCg is calculated area-weighted in the classical way and assumes that surfaces are not overlapping. In a well built model, AbsC and AbsCg should be very close.

Early/late energy-ratio based parameters Early-to-late energy-ratio based parameters (or measures) such as Deutlichkeit (D-50), Clarity (C-80) etc. are calculated according to their respective definitions [Cremer,ISO].

138


Lateral measures LF and LFC 80

LFC = 100

∫

ms

h 2 (t ) | cos(θ ) | dt

5ms 80

(2.5-3)

ms 2

∫h

(t ) dt

0

where

θ

reflection angle related to the ear to ear axis on an assumed listener looking as defined by the head-direction in General settings. [ISO] states that the headdirection should be towards the source. 2

LF uses cos (θ ) instead. LFC is better correlated to the spatial impression but LF is easier to measure [Barron-81].

Strength (G) G is the sound pressure level relative the free field sound pressure at 10 m from the source (assuming an omni-directional source with the same output power as the actual source):

( )

( )

2 2 G = 10 log ptot − 10 log p10

(2.5-4)

where p10

sound pressure rms for the direct sound at 10 m distance in free field, Pa

ptot

total sound pressure rms, Pa

Multiple source G estimates

If more than one source is used (Audience area mapping or Source addition), G is calculated as:         G = 10 log  (2.5-5) p2i / pi210  = 10 log  10Gi 10         

∑

∑

where

pI

total rms pressure squared at a receiver or map point caused by source i

pi10

rms pressure squared at 10 m in free field for an omni source with the same total output power as the real source i

Gi

G for source i

G is not really defined other than for a single source but better to include something reasonable than not to show a requested plot. One can say that G for multiple sources normalizes so that the level is independent of individual source power, as if all sources were identical.

139


Early Decay-Time (EDT) EDT is the decay time evaluated from 0 to -10 dB on the backward integrated echogram. Due to difficulties of doing a least square fit to the sometimes not very linear early backward integrated curve this procedure, as suggested in draft standards for measurements, has its problems (especially with a very strong direct sound). The best way to determine EDT remains to be defined, see e.g. [Bradley-96].

RASTI RApid Speech Transition Index is calculated from the echogram according to the procedures described in [Houtgast-80] and [Jacobsen-86]. Background noise levels can be specified in the General settings/Acoustic environment dialog. Rather than using generic speech levels for the 500 Hz and 2 kHz octave-bands (with a 9 dB difference), the actual levels set by Lp1m in the source-file are used. It is the user's own responsibility to set these to appropriate levels. Predefined constants or Lp_voice shortcuts can be used for this purpose, see source-file syntax section 2.3. No RASTI values will be given if the SPL is more then 10 dB Below the background noise level.

STI Speech Transition Index is like RASTI calculated from the echogram. Original STI is based on 14 modulation frequencies (0.63 - 12.5 Hz) and the octave-bands 125 - 8k Hz but fewer bands and/or different weights can be selected by the user in Preferences. One available option is to use "STI modified" based on the Articulation Index (AI) weights from French & Steinberg (AI is a predecessor to STI but took into account only the S/N ratio in each band and no modulation, see [Steeneken-67]. It appears that MLSSA [MLSSA] is the only measurement system where these weights are used as an option. Further options are from CEI/IEC 60286-16, 2003-05 where also the redundancy correction (beta) factors are included. Background noise levels can be specified in General settings/Acoustic environment. Rather than using generic speech levels, the actual levels set by Lp1m in the source-file are used. It is the user's own responsibility to set these to appropriate speech levels. Predefined constants or Lp_voice_IEC shortcuts can be used for this purpose, see source-file. No STI values will be given if the SPL is more then 10 dB below the background noise level.

Normalized values It can sometimes be useful to compare estimated (or measured) parameters to values expected from an exponential decay calculated from the reverberation time [Cremer, p434]. It is simply a matter of using a pure exponential function in the integrals involved:

E (t ) = E 0 e −δ t

(2.5-6)

13.816 T

(2.5-7)

δ = where T

reverberation time, s

The values thus obtained do, however, not take into account the direct sound and can only work far away from the source where the reverberant field dominates. Barron and Lee therefore developed a new, but still simple, theory for expected values in concert halls [Barron88]. Theoretical calculations where compared with measurements in 15 different concert halls. The new theory matched measured values much better. The formulas used were:

140

2. PREDICTION MODULE For the direct sound:

2 = pdir

W ΓZ 0

(2.5-8)

4π r 2

where W

total output power of the source, W

Γ

directivity factor (not included originally)

Z0

wave impedance in air, kg s /m

r

distance from source to receiver, m.

2

For the subsequent decay:

p 2 (t ) =

τ=

4 W Z 0 e -δ t e -δτ A

(2.5-9)

r c0

(2.5-10)

where 2

absorption area, m Sabine

A

The directivity factor was not originally included in the theory as the measurements were performed using an omni-directional source, but with computerized prediction also the directivity can be taken into account. The normalization can be switched on in the Full detailed calculation dialog. The parameters G, D-50, C-80, and SPL will then have a line each (in E_ss_rr.TXT and PARAM_ss.TXT) where the expected values have been subtracted. A positive value means that the predicted value is higher than the expected. A plot-file (NTREND_ss.PLT) similar to TREND_ss.PLT is also created if Normalized trend is selected.

Echo-disturbance (EKgrad) The Full detailed calculation dialog has an option for an echo-disturbance criterion. The calculated graphs are based on an article by [Dietsch&Kraak-86] and its application in [Roy&Gimbott-93]. It is too much to go into the detailed background of EKgrad here, users are recommended to study the article, this documentation will only give a shorter version. The general hypothesis behind this criterion is that an echo disturbance is given by:

t d pd

n

= const

(2.5-11)

where td

is the arrival time of a reflection (after the direct sound), s

pd

total pressure of the reflection, Pa

n

an exponent to be determined by listening tests

141

2. PREDICTION MODULE i.e. an early reflection has to be strong to give a certain disturbance while for a late reflection the level can be less. The result of the listening test resulted in a sliding evaluation of centertime. To better apply to the hearing, it was not based on pressure squared as for ISO Ts (i.e. n = 2) but on the absolute pressure |p| for music (n = 1) and |p|^2/3 for speech (n = 2/3). The procedure is to first evaluate this special Ts as a function of time: τ

∫ t p(t )

n

dt

0

t s (τ ) = τ

∫ p(t )

(2.5-12) n

dt

0

This function is then smoothed with a 9 ms window for speech and a 14 ms window for music. The resulting two functions of time are differentiated and plotted (i.e. fast increases in the Ts functions give high values and risks for echo disturbances). From listening tests criteria were determined for 10% and 50% disturbance and indicates the risk for audible echoes. The authors also indicate that if the EKgrad exceeds the limits at times separated by more than 50 ms (speech) and 80 ms (music) it indicates a risk for audible multiple echoes, see table below for summary of all constants.

EKgrad constants

Speech

Music

Exponent n

2/3

1

Smoothing

9 ms

14 ms

EKgrad,10%

0.9

1.5

EKgrad,50%

1.0

1.8

Multiple echo risk

50 ms

80 ms

Note: The predicted curves may indicate echo risks also for reflections arriving before 50 ms and that is highly unlikely. Fig. 11 in the D&K article indicates that predicted and measured disturbance differ for < 50 ms reflection delays indicating that the method works best for > 50 ms delays. Strong reflections, or sound arrival from other sources in a PA, between ca. 20-50 ms can create coloration so they may still be important to consider, but they will typically not be interpreted as echoes.

Stage support The Stage support parameters (STearly, STlate, STtotal) [Gade-92] are calculated for any receivers that are placed at 1 m (±1 cm) from the source and at 1 m (± 1 cm) height above the closest surface below the source. See also [O'Keefe] for an overview of stage acoustics parameters. Note 1: Stage support assumes an omni source and if that is not the case a warning is given along with the predicted values. The reason is that it could be interesting to use an actual measured dodecahedron source that will not be ideally omni but may be useful to compare to measured data using the same source. Note 2: the source is assumed to be placed no closer than 4 m from a wall, this is up to the user to ensure. For further details, see [Gade-92].

142


Echogram calculation All echograms presented begin at the arrival of the direct sound (marked by a small ring on echogram graphs if the source is visible from the receiver). For receivers where the source is invisible due to blocking walls or balconies, the echogram and thereby parameter calculations still starts at the time the direct sound would have arrived had the source been visible. Note that most parameters loose their meaning if the direct sound is not present and predictions become less accurate. Echogram graphs

Echogram graphs come in a variety of forms. A forward integrated echogram approximates:

t

F (t ) =

∫h

2

(τ ) dτ

(2.5-13)

0

where

2

h (t)

2

impulse response squared, Pa

Shown in E_ss_rr_oct.PLT (Full detailed calculation) and in I_ss_rr_oct.PLT (Early part detailed ISM, option). A backward integrated echogram approximates: ∞

∫

B(t ) = h 2 (τ ) dτ

(2.5-14)

t

Shown in E_ss_rr_oct.PLT, E_ss_rr_INT.PLT (Full detailed calculation) and in I_ss_rr_oct.PLT (Early part detailed ISM, option). The crossing of the forward and backward integrated curve will be an estimate of "Rise time" as defined by W.J. Jordan [Cremer, p 433]. Although the time-constant of the ear is not easily defined, it is often useful to view echograms convolved or smoothed with some kind of filter function, e.g. exponential (low-pass), so that the effects of reflection gaps and/or many reflections arriving close in time can be studied in the echogram:

M (t ) = h2 (t ) ∗ G (t )

(2.5-15)

where G(t) ∗

filter-function (exponential/triangular/rectangular) convolution symbol

Shown in E_ss_rr_oct.PLT (Full detailed calculation, filter-type/time-constant option). By using a linear scale of Sone and also taking the ear integration into account, a function more like how the ear hears it can be plotted. The function used for impulse sound is the square root of h. See [Cremer, pp 456-464] and [Cremer-89]. The ear-simulated echogram approximates:

143

2. PREDICTION MODULE 1/ 4

 ∞ h 2 (τ ) e - (t -τ )/ε dτ  S (t ) =  ∫ t 

(2.5-16)

where S(t) ε

ear-simulated function, Sone ear integration time constant, s

Shown in I_ss_rr_oct.PLT (Early part detailed ISM, option).

Directivity handling The source directivity handling depends on the type of directivity selected. For the types SD0, SD1, CF1 and CF2 the application is straightforward since they all have the value at the source axis as the reference (set to 0 dB in the directivity balloon). For the SD2 type however, based on the DLL Directivity Interface (DDI), a particular model may have a distance dependent directivity (typically arrays). With SD2, the direct sound and first order reflections are taking into account the distance dependence while for higher order reflections the far-field directivity is used. For details on the various directivity formats, turn to the Directivity Module section 4. A note on arrays: If the physical dimensions of an array modeled in a DDI interface is considerable, there is a risk of errors since the rays etc. originate from only one point in the prediction (the acoustical center of the source; the reference point for an array) but with the directivity of the full array. Consider the two examples below in Fig. 2.12 and 2.13. In these cases the array will in reality not give the predicted directivity at the receiver because some transducers may be blocked (for higher order reflections it is further complicated). However, an array is seldom placed so that its direct sound is screened and with a long array is also associated a strong directivity meaning that the direct sound will dominate so this error is likely to be of limited importance. The ability to model the distance dependent directivity while still handling the source as a single entity is of greater importance than any errors of this kind.

Actual array receiver

SD2 source Fig. 2.12 Direct sound from SD2 arrays: some direct element paths are cut off.

144

2. PREDICTION MODULE Image array

Wall A

Actual array receiver Wall B

Image array

Fig. 2.13 Reflected sound from SD2 arrays: some reflected element paths are cut off.

Semi-transparency Each of the prediction methods handles semi-transparency slightly different since it is difficult to implement it exactly the same in all cases. The basic procedure is however as follows: The direct sound goes (deterministically) through a maximum of one semitransparent surface and it is then attenuated by (1 - α)τ. st

1 order reflections are correspondingly attenuated by (1 - α)(1 - τ). Higher order transmission is handled by a random process, see below, see Fig. 2.14. The sum of the three components must be = 1 so that the energy is conserved:

α (1 - α) τ

transmitted

(1 - α)(1 - τ)

reflected

α

absorbed

(1- α) τ + (1- α)(1 - τ) + α

=1

1

(1-α) τ

(1- α)(1- τ)

Fig. 2.14 Relations between absorbed, transmitted and reflected energy. nd

2 order and higher transmission is random depending on the transmission coefficient. Setting α = 1.0 (100%) means a totally absorbing surface and there can be no transmission, while setting τ = 1.0 (100%) will result in a material that has no reflection but only absorption (much like air absorption). Neither of these selections are meaningful for semi-transparency. It is not obvious how to select α and τ since many materials such as cloths are measured hanging against a hard wall and the resulting absorption coefficient is not the proper one to

145


use here. To be able to estimate the transmission of e.g. cloths, see rererences [Ingaard], [WinFlag] or [Cremer-2]. Sabine and other classical measures loose their meaning when semi-transparent materials are used since coupled room systems may be created, so be careful not to over-interpret them. Note on τ: what is available from measurements is not necessarily what is suitable for use with prediction. Consider a measured reduction index, R, defined as: R = -10 log(W in/W tr)

(2.5-17)

where W in is the incident power and W tr is the transmitted power. When measuring W tr the effects of both absorption in the structure and reflection due to impedance mismatch is included and they are hard to separate. Expressed in the quantities above it means: R = -10 log(W in/W tr) = -10 log((1 - α) τ)

(2.5-18)

giving the τ values required for the semi-transparency to be calculated from R as: -R/10

τ = 10

/ (1 - α)

(2.5-19)

Note on α: the α needed is that due to pure dissipation (heat) rather than the normal absorption coefficient that does not distinguish between what is absorbed and what is transmitted. The AbsTex software [Ognedal] may be used to calculate the values required, contact CATT for details.

Calculation recommendations Audience area mapping At least the number of rays estimated by the Auto number setting. If warnings are on (File|Preferences) a warning will be given if the selected number of rays is less than half of what would have been estimated by the Auto number setting. A ray truncation time corresponding about -30 dB (50% of the RT).

Early part detailed ISM No definite guidelines can be given here since it depends on what type of information is wanted. For example, to check for potential late strong echoes a long truncation time should be used but orders higher than 3 would in practice be diffused and would seldom cause echoes.

Full detailed calculation At least the number of rays estimated by the Auto number setting. If warnings are on (File|Preferences) a warning will be given if the selected number of rays is less than half of what would have been estimated by the Auto number setting. A ray truncation time corresponding about -60 dB (100% of the RT). The longer truncation time as compare to mapping is because the T-30 estimate.

146


Diffuse reflection Diffuse reflection heavily influences the RT estimate and many other parameters. It is very important to give reasonable estimates for the coefficients. ·

a minimum of 20% default diffusion is recommended for average-size flat smooth surfaces (10% for big flat smooth surfaces)

·

30 to 80% for the audience (30% for 125 Hz and increasing up to 80% at 4 kHz).

·

for rough surfaces a high value (80%) should be assigned where the roughness scale is of the order of the wavelength (i.e. if the roughness is about 3 dm the 1 kHz diffusion should be set high). If in doubt, it is better to assign too high than too low scattering coefficients. The GEO-function estimate() can be used instead of giving values

·

automatic edge diffusion can be applied to reflectors, windows, cupboards, tables etc. that often are flat and smooth and only give diffusion if their size is small in relation to the wavelength.

·

with the advent of an ISO standard for measuring scattering coefficients guidelines [ISOscatt] measurements will gradually be available as guidelines but in the majority of rooms most scattering coefficients will have to be estimated.

2.6 Tips and Tricks This section contains a number of recommendations and examples on how to get the most out of CATT-Acoustic. ·

It is strongly recommended not to make too detailed models but to rather model with less detail and use a frequency dependent scattering coefficient to compensate for detail. This is due to how geometrical acoustics works and affording the use of more rays with a simpler model is highly likely to give better results than affording fewer rays with a more detailed one, in fact even using the same number of rays for both, the less detailed model is likely to give better prediction results especially at the low end.

·

Work in steps - first build the main parts of the hall then the details. View the geometry as often as possible between changes. Don’t add too much new geometry between each viewing.

·

Use constants (LOCAL, GLOBAL) - it is much easier to make changes. Use constants especially where you suspect changes. It might seem to be a great effort to plan the use of constants and to do other structuring of the geometry but this initial effort almost always pays off in the long run.

·

If one corner's z-value (for example) always has the same z-value as another corner use the z() function (or x(), y() for x,y-values):

10 -3 ... 20 -7

5

12.5

12

z(10)

This signals that corner 20 does not just happen to have the same z as corner 10 but that it is always the same z and should automatically follow any changes in corner 10. This makes the model both simpler to build, change and interpret. ·

To prevent warped plane use the lock() tool whenever possible, typically for slanted floors or ceilings but also for walls.

·

The x(), y(), z() functions can be used to create a copy of a part of the model, which then can be treated as an object and be "lifted off".

147


·

The COPY function can be used to create multiple identical reflectors. As the structure of the reflector is defined once, the structure can, initially, be made very simple (typically rectangular) for rough tests and then, towards the final calculation, be changed to a more realistic shape. If the coadd and pladd values, from start, are chosen to accommodate for the realistic shape, all that is needed is to create the shape (around the same origin) and the result will be instant multiple realistically-shaped reflectors.

·

The possibility to define one surface property to equal another (ABS name1 = name2) can be used for fast and secure absorption changes. If e.g. all parts of the ceiling that will have the same absorption are assigned e.g. ceilingabs, then ceilingabs can be defined as e.g. wood and when changed to e.g. plaster all ceiling parts will change absorption. This works in a similar manner to "styles" in a word-processor.

·

With symmetrical designs where the MIRROR directive can be used, add the ceiling and other “bridging” planes as single planes instead of utilizing the mirroring to avoid potentially dangerous edges in the middle of the ceiling. Since the mirror function creates all corners it is just a matter of defining the plane and it can be defined in the same file since planes bridging over from the negative x side to the positive x side are not mirrored. If all parts are mirrored more planes are generated which increases the calculation time and also makes wire-frame plots harder to interpret.

·

INCLUDE GEO-files - especially for parts of the hall you would like to compare with and without. It is also possible to include alternative versions of e.g. ceiling reflectors and compare them without having to make a complete new version of the hall. Use constructs like:

GLOBAL highbalcony = TRUE IF highbalcony THEN INCLUDE highbalc.GEO ELSE INCLUDE lowbalc.GEO ENDIF ·

Use a subset of planes (General settings/(Plane use)) or disable INCLUDEs by turning them into comments when you build the model. This makes geometry plots and the debugfile easier to interpret as you can view/check parts instead of the whole hall. With a symmetrical hall and use of the MIRROR directive, all mirrored planes can be switched off while building the model: if planes are mirrored by 500, switch off all planes e.g. 501-999. This is better than to temporarily comment out the MIRROR directive since also nonmirrored planes, such as floors bridging from -x to +x, will be included in the design and plots, assuming they have plane ids ≤ 500).

·

Always define reflectors as objects. They are always separate from the main geometry and are usually moved and rotated. Leave the work of calculating the corner coordinates to the computer! If corners coordinates actually are needed - generate a debug file and read the corresponding values from the corner list. If multiple identical reflectors are intended the COPY directive will simplify the job.

·

Simulate audience areas by a plane at about shoulder height (about 0.8 m) having the normal pointing up. The audience should be "walled-in" by planes and the floor below should not be modeled, but if they are not (see Fig. 2.9), possibly for simpler runs for early reflections or direct sound mapping, the back-side of the audience plane must also be defined and assigned the pre-defined absorption TOTABS. The reason for this is that this plane will then catch some impossible rays running under the audience. The audience should typically have a high degree of diffusing properties assigned to it (30 to 80% ascending from 125 Hz to 4kHz).

·

If the hall has audience areas the lock() tool can be used to set the receiver positions at an exact height over the audience plane. First, create three help corners at a fixed height above three audience plane corners. Then lock() the z-coordinate of all receiver positions to a plane defined by these three corners.

·

Typically use the "sum" or "1k" for octave-band plot-file output. Only select all bands if you

148


really want a detailed check (e.g. for a certain receiver position) or a lot of plot-files will be generated. If all plot-file options are used together with many receivers and all octavebands several hundreds of plot-files may be created! This may seem strange but CATTAcoustic is built to give the possibility of a lot of information - but if you don't want/need it you don't have to request it - at least not for all receivers! · Reminders about the source spectrum: If a "white spectrum" on the source axis is desired the values should increase by 3 dB/octave (or use the predefined Lp_white). The octaveband “sum” value of the acoustic parameters, D-50 etc., RASTI and STI are affected by the source spectrum as are Source additions. If a comparison between measurements and calculations in a certain hall is desired and octave-band values cannot be extracted from the measurements, be sure to match source spectrum to the actual source or the comparison may be meaningless. ·

To create a soffit- or ceiling-mounted source the source position must not be placed exactly inside a wall or ceiling but slightly in front of it and an absorbing patch must be placed behind it on the wall (unless the directivity is modelled as half-space radiating). Soffit-mounted sources are difficult to handle since the directivity of a source when placed in a large baffle changes from that of the free source. If a directivity measured in a large baffle is not available a free-field directivity must be used with great caution.

·

To check for occurrence of late low-order reflections, that might cause echoes, do a lowerorder run with specular order 2-3, a long early part truncation time. The history plot-file (H_ss_rr_oct.PLT) will give the echo information.

·

Avoid placing both source and receivers exactly on the symmetry-line of a model. Same goes for measurements since it may introduce interference effects.

149


2.7 Trouble-shooting This section lists probable causes to Prediction module errors.

Geometrical errors The program performs extensive checking on the consistency of the geometrical model. In spite of this checking, some geometrical errors might be so severe that the program crashes and even, occasionally, dead-locks the computer. Such errors will typically occur while the program is calculating plane information. Below are some probable causes listed: ·

you have happened to "cross" a plane by a mistake in the entering order of the plane corners (an example: what was intended to be a rectangular plane has been entered as a "bow tie"). See Fig. 2.15. 1

2

1

3 correct / 1 2 3 4 /

3

4

2

4 wrong / 1 4 2 3 /

Fig. 2.15 Plane corner entering order error

·

you have happened to make an edge of e.g. a reflector cut another plane. You should not put a corner of a plane in the "middle" of another plane without using that corner in the definition of both planes (see Fig. 2.5). By using immediate debug (Geometry view/check dialog) errors of this kind will be trapped while building the model.

·

you have happened to make an error in a corner coordinate causing one or some of the corners to be "out of the plane" (i.e. a warped plane) when a plane equation is built from three of the corners. By using immediate debug (Geometry view/check dialog) large errors of this kind will be trapped while building the model but smaller will pass and can cause an error. the DEBUG.TXT file (Geometry view/check dialog) will always list all corners even for planes not used and you can check if there is an error in coordinate values or entering order.

·

when using SCALE together with external coordinates (defined in another GEO-file) by using e.g. x(), a value may be scaled twice if the coordinate was scaled in the file it was defined in too. In such cases use e.g. x(120)/s, if 120 is a corner defined in another file with a global x scale-factor of s, to inhibit the extra scaling.

·

if the volume or Interactive RT estimation is impossible it is due to actual holes (i.e. an actually open model), overlapping planes or inaccurate (warped) planes.

The problem plane id is often listed in the left status bar when an error occurs but it depends on the error. One way to find a problem plane is to insert a RETURN statement in a PLANES section and gradually move it down until the error occurs, thus identifying the problem plane. For most efficient search, use a binary search technique so that the return statement is first put in the middle of the PLANES section, thus identifying if the problem is in the first or last half of the planes defined, and then moved to the middle of the problem half etc. However, best is to not add too many planes before viewing the model. Geometrical errors will confuse the algorithms and will result in calculation errors. There is no

150


way of knowing for sure if the error will be of great importance or not - the geometry should be as accurate as reasonably possible. Even if a model is simple it should be correct!

A good way to summarize the modeling process for realistic results is: Model all surfaces that can be reached by sound but don’t model any surface that can’t be reached by sound!

It also helps remembering that the model to be made is an acoustical one and not a visual one. Another one-liner that can help in understanding how a closed model must be is (here it is assume that a ray is not stopped by a plane): A ray from any position inside a room, sent out in any direction, must pass an odd number of planes on its way to infinity

For example, with a simple shoebox a ray will always just pass one plane (=odd number) while if there is properly modeled reflector inside, a ray may pass three planes (=odd number) while if the reflector is made just of one plane a ray may pass only two planes (=even=wrong modeling), see fig 2.16. 3

1

2

2 1 correct

correct

1 wrong

Fig. 2.16 Modeling all surfaces that can be reached by sound.

Audience area mapping problems ·

If boxes are placed on top of a floor, defined as an audience plane, and with a big Map step it may look like positions inside the box are hit. This is simply because some part of the receiver spheres are outside of the box.

Early part calculation problems ·

-

Full detailed calculation problems ·

If you obtain strange results or numerical errors one probable cause is that you have created an acoustically impossible or unreasonable model e.g. by using the predefined absorption TOTABS carelessly or defined another absorption property that has 99.9% absorption for all or some octaves. By using such non-realistic absorption values, the early part might be possible to interpret but the late part will be very strange since rays

151


encountering a totally absorbing surface will be truncated. If you want to test a "very high absorption" use 95-99% instead. The 100% absorption is reserved for special purposes such as under the audience surface as described in section 2.6 but is best avoided. Note that measured absorption giving 100% (or even more) is due to a limitation in the measurement method where a very high absorption destroys the diffuse field (that is assumed when deriving the coefficients). A small example illustrating the danger of using totally absorbing surfaces: Consider one reflection on a 99% absorber followed by 5 successive reflections on a 25% absorber. The 5 reflection is then attenuated by a factor (1-0.99)(1-0.25) = 0.0024. If there instead were 6 6 successive reflections on the same type of absorber it would correspond to (1-α) = 0.0024 ⇒ α = 0.63 i.e. not an absorber that would be approximated by total absorption. If TOTABS were used instead of 99% the ray would be truncated while with 99% there is a chance that the high absorption of one surface is compensated for by several subsequent hits on hard surfaces. ·

If reverberation decay curves become very non-linear (E_s_rr_oct.PLT and E_ss_rr_INT.PLT) and thereby T-30 very different from T-15 and EyrT the most probable causes are a lack of diffuse reflection or that you have created a model with TOTABS planes or that contains holes. In both types of output text-files (E_ss_rr.TXT and PARAM_ss.TXT) and in the reverberation time overview plot-file (RT_ss.PLT) the number of rays lost and/or absorbed as compared to the number of rays sent out can be seen. If more then 10% are lost a warning is given. Lost rays are caused by holes in the model, inaccurate planes and by the finite numerical accuracy. Absorbed rays are caused by use of the predefined TOTABS absorption.

152

3. SURFACE PROPERTIES MODULE

3. SURFACE PROPERTIES MODULE The Surface properties module manages an absorption/scattering coefficient library contained in the Surface properties library file as selected in Preferences. The library files (*.DAT in the CATTDATA-folder ABSLIBS) are in a private binary format but text-file import/export is provided. The library file is implemented with variable-length records so it is only the records that actually use long descriptions and/or references that occupy space in the library which means a flexible but compact format. This module can also create a new empty library or a copy of an existing library that then can be extended. Absorption/scattering properties declared in the library will be known in every GEO-file. In addition to using the library, properties can be defined directly in a GEO-file and an attempt to re-declare a property will be trapped, see the ABS directive, Section 2.3. Note that properties with no diffusion defined may still be assigned a default diffusion by selecting Prediction:General settings/Surface default. It is also possible to complement an absorption-only library property with scattering coefficients in the GEO-file using the ABS directive. For predictions, the Surface properties module does not have to be open, but they will run a fraction faster since the library-file does not have to be loaded each time.

3.1 Menus File|Save library

Saves the current library to disk. Before saving, the current version on disk is backed up to a file with the same name but with extension .~DA. File|Import|CATT Exchange TXT...

Imports an absorption/diffusion properties library in the extended text format intended for library exchange and updates. The syntax is the same as used for the GEO-file directive (ABS) in section 2.3 except that the ABS keyword is left out and that three extra mandatory fields per record are added. If a key name already exists it may, at import, either replace the current one or be skipped. The file has the format: absname = <α125... α4k [: α8k α16k]> m [m_data] [{R G B}]

C: category D: description R: reference ... Where the scattering coefficient part, 8k and 16k values and the RGB color are optional. m_data is currently not used. The C:, D: and R: fields may be empty but the field labels must be present. Property names can be in mixed case but will always be converted to uppercase. CATT-Edit can create an empty record template if some values are entered manually.

153


If a category does not exist it will be created and if no category is specified, category "None" is assigned. Comments are introduced by ";" and blank lines are allowed Example exchange-file with three properties:

CONCRETE = < 1 1 1 2 3 4 > C: walls D: painted R: Building research data HARDWOOD = < 12 10 8 7 6 5 > {200 100 0} C: walls D: 15 mm thick R: estimated values DIFFUSOR_1 = <12 10 10 10 10 10> L <30 40 50 60 70 80 > C:diffusors D:hand-made in wood R:estimated values File|Import |CATT v5.0 TXT...

Imports a text-file in the older CATT-Acoustic v5.0 for DOS absorption library format. The syntax is the same as used for the GEO-file directive (ABS) in section 2.3 except that the ABS keyword is left out. The category will be "None" and description/reference fields will be empty. If a key name already exists it may, at import, either replace the current one or be skipped. A better alternative than to import an old-format file directly, is to first load the old library into CATT-Edit or another suitable editor, convert all comments etc. to the Windows ANSI character set, and then transform comments to be descriptions or references as appropriate. CATT-Edit can also assist with a new-format template. The new file format can then be imported as CATT exchange text, see below where also the format is described. The file has the format:

absname = <α125...α4k> [ m ] ... Where the scattering coefficient part is optional. Comments introduced by ";" and blank lines are allowed. Example file with three properties:

CONCRETE = < 1 1 1 2 3 4 > HARDWOOD = < 5 5 7 7 7 8 > DIFFUSOR_1 = <12 10 10 10 10 10> L <30 40 50 60 70 80 > File|Export CATT Exchange TXT

Exports the properties library in the extended text format, see above. This is intended for library data exchange and updates, and possibly manual editing since it is sometimes faster to work directly in a text editor. Edit|Copy Key to Clipboard

Places the property name (e.g. "WOOD") on the Clipboard so that it can be pasted into a GEOfile. This is the same as double-clicking on a property name in the list except that an open Prediction:Interactive RT estimate dialog will not be updated with the new property. Note that

154


an absorption-only property (say "WOOD") can be complemented by scattering coefficients when used in a GEO-file: ABS MyWOOD = WOOD L <5 10 15 20 25 30> if the actual wood construction is not flat. Edit|Copy Definition to Clipboard

Places the complete property definition (e.g. "ABS WOOD = <12 10 8 7 6 5>") on the Clipboard so that it can be pasted into a GEO-file and be renamed and edited e.g. to "ABS MYWOOD = <12 10 8 7 6 5> L <10 20 30 30 20 20>". Utilities|Create a New Library As...

Creates a new generic library that can be selected in Preferences and be filled in with more properties. Utilities|Save Copy of Library As...

Creates a new library as a copy of the current one. The new library can be selected in Preferences and be filled in with more properties. Category|New...

Creates a new category.

Category|Remove... / Category|Rename...

Displays a list of all current categories and the selected one can be removed/renamed. All properties in the removed category will be assigned the category "None".

Options|Sort Order...

Sorts the entries according to various criteria.

155


Key, Category, Reference

Sorts alphabetically according to Windows ANSI character set. 125/250 Alpha , 500/1k Alpha, 2k/4k Alpha

Sorts according to the type of absorber. The sorting function used is:

(α( f ) + α( f ) ) / ∑ α (f ) 2

1

i

2

i

where f1 = 125 and f2 = 250 for 125/250 Alpha etc.

This means that with e.g. 125/250 Alpha a dedicated low-frequency absorber is sorted before a wide-band absorber with the same values at f1 and f2. 125-4k Alpha

Sorts according to average absorption coefficient (125 – 4k Hz). 125-4k Diffusion

Sorts according to average total scattering coefficient (125 – 4k Hz). For all sorting options entries are sub-sorted after Key.

156


3.2 Dialogs The main dialog displays a scrollable list-box showing all defined properties. When a property is selected the graphs and other related information about the property is updated.

List

Lists all defined properties according to the Options|Sort Order menu. The current sort order is displayed in the module window title. Search key

The key or name of the property. This key is the name that is used in GEO-files. To use a property in a GEO-file without having to type it in, use any of the following:

⋅

double-click the property in the list-box

⋅

press Enter when a property is selected

⋅

select Edit|Copy Key To Clipboard

The property name is then transferred to the clipboard and can be pasted into a GEO-file using the CATT-Edit Edit|Paste menu. A similar procedure can be used when assigning surface properties in the AutoCAD interface. Property key names are always converted to upper-case and may consist of a maximum of 15 characters being A-Z, _, 0-9, and native language characters. A key must always start with a letter. Note: there are two predefined absorption properties: TOTREF and TOTABS that do not appear in the library. TOTABS is an artificial total absorption used for the back side of "floating" audience planes and similar planes for quick tests but should be used very seldom (better to use 99,9%).

157


Category

A property category e.g. Walls, Windows, Audience etc. See the Category menu for adding, removing and renaming categories. Description

A description of the selected property such as “13 mm plaster-board on studs” etc. Color

A color assigned to the property. Reference

Source of the data. This is an important information since properties are of limited value if it is not known how they are measured. Add/Update

Updates (if already present) or adds to the list if a new Search key has been entered. A property can be changed either by entering values in % in the edit-controls below the graphs or by direct manipulation of the small handles attached to the curves. If Scattering is selected as "None", the scattering coefficients still remain in the record so when turned back to Lambert they appear as before. There is no special "New property" button. To add a new property simply change the key and the coefficients as required and select the Add/Update button. If the key name already exists you will be prompted for the action to take. In this manner it is very handy to add several properties from the same source since the reference field needs to be filled in only once.

158

4. DIRECTIVITY MODULE

4. DIRECTIVITY MODULE The source Directivity module allows graphical and numerical editing of source directivity library-files (native CATT *.SD0, *.SD1, *.SD2) and viewing the CLF format (*.CF1, *.CF2 [CLF]) in the CATTDATA-folder SD or in sub-folders to SD - one level deep). The directivity key-name (i.e. the file-name including extension) is entered in a source-file at prediction. If directivity files are located in an SD sub-folder the full name of is sub-folder\name.SDx or sub-folder\name.CFx. "OMNI" is a pre-defined natural omni-directional source and need not have a directivity-file. Five formats are available, a 15° polar-interpolated format (SD0), a 10° full-space measured format (SD1), The CLF formats (CF1,CF2) and a general format capable of very high resolution and distance dependence based on a DLL Directivity Interface, DDI (SD2).

4.1 Menus File|Open From Collection...

Displays a dialog where directivities from defined collections can be selected. A collection is a list of directivities but does not contain the directivity files themselves.

Help (Collection)

Select from collections

Displays collection-specific help if available (CHM, HLP, PDF, HTM or TXT) File|Edit Collections...

Displays a dialog where directivity collections can be created and edited. Typical use of collections is to group directivities into manufacturer groups but a user can also make his own collections of often used or "favourite" directivities. New (Collection)

Edit collections

159


Creates a new collection (stored as .COL files in CATTDATA-folder SD). Delete (Collection)

Edit collections

Deletes the selected collection. Note: the directivity files themselves are not deleted, only the collection. Help (Collection)

Edit collections

Display collection-specific help if available (CHM, HLP, PDF, HTM or TXT). Add... (Directivity)

Edit collections

Adds a new directivity to the currently selected collection. Remove (Collection)

Edit collections

Removes a directivity from the currently selected collection. Note: the directivity file itself is not deleted, only the reference to it in the collection. File|New Directivity|15° format (SD0)

Creates new generic data for the 15° polar-interpolated format. File|New Directivity|10° format (SD1)

Creates new generic data for the 10° full-space format. File|New Directivity|DLL-format (SD2)

Creates new default data for the DLL-interface format. Displays a dialog listing all installed Brand DLLs and the models they implement.

Each brand DLL is installed in a CATTDATA-folder SD2DATA sub-folder with the same name as the DLL itself and will be automatically recognized. Optional dedicated help is included. Example for the CATT_Generic module:

160

4. DIRECTIVITY MODULE CATTDATA-folder SD2DATA (folder) CATT_Generic (folder) CATT_Generic.DLL CATT_Generic.CHM File|Open Directivity|CLF format (CF1,CF2)...

Opens a CLF-format directivity file from the CATTDATA-folder SD or sub-folders [CLF]. Note: CLF-format files cannot be edited or exported. File|Open Directivity|15° format (SD0)...

Opens a 15° interpolated format directivity file from the CATTDATA-folder SD or sub-folders. File|Open Directivity|10° format (SD1)...

Opens a 10° full-space format directivity file from the CATTDATA-folder SD or sub-folders. File|Open Directivity|DLL- format (SD2)...

Opens a DLL-interface format file from the CATTDATA-folder SD or sub-folders. An SD2-file is normally only edited via the interface but it is in text-format and can be loaded into an editor for more efficient editing of e.g. complex array definitions: Syntax (keywords in BOLD, values in italic, explanations in plain text, [ ] items are optional): CATT-SD2 Version = file-version Model = model-name Accuracy = level Description = description Visuals = none|extents|marker (version 1 only) Brand = dll-name DLLversion = dll-version lines of optional model dependent data CATT-SD2

a file-type tag

Version

a file-format version number, 0 for v7.1 and v7.2, 1 for v8.0

Model

a model name as defined in the DLL help.

Accuracy

some models may have optional accuracy levels (see DLL help). If not this value should be 1.

Description

a descriptive name of the source

Visuals

for version 1 only: If marker then the DLL returns marker-files for all array elements.

Brand

name of the DLL (without .DLL) that handles the model. An SD2 DLL is placed in the CATTDATA-folder SD2DATA sub-folder with the same name as the DLL (assuming that Brand = BrandX, the DLL will be CATTDATA-folder\SD2DATA\BrandX\BrandX.DLL).

DLLversion

the optional model-dependent data assumes a DLL with this version.

161

4. DIRECTIVITY MODULE lines of optional model dependent data any number of lines of data that each DLL knows how to interpret.

The model-dependent optional data can be edited via the SD2 dialog but for some models it may be more efficient to use an editor to edit the SD2-file directly. An SD2-file has optional dedicated help on each model implemented (Help button in the SD2 dialog). An example SD2file: CATT-SD2 VERSION = 1 MODEL = Intellivox-2B ACCURACY = 3 DESCRIPTION = Duran column array VISUALS = marker BRAND = Duran_CATT DLLVERSION = 4 OPENINGANGLE = 10 FOCUSDISTANCE = 20 AZIMUTH = 0 CALIBRATED = yes

File|Open Directivity|Any Format...

Opens any supported format (SD0, SD1, SD2, CF1 or CF2 and possibly future formats). File|Save Directivity

Saves the directivity to the current file. File|Save Directivity As...

Saves the directivity to a new file of the same type in the CATTDATA sub-folder SD folder or in sub-folders to SD (one level deep). The file-name can have max. 15 characters including the sub-folder name and \ if saved in a sub-folder. Source directivity file names can use hyphens (-), underscores (_), letters and numbers. If a file is saved in a sub-folder it may start with a number so that BRAND\modelnumber can be used. Note: CLF files cannot be saved. File|Import Directivity|CATT-SD0 TXT...

Imports the 15° interpolated format. In the SD0 text-format the source directivity is entered as polar measurements for horizontal and vertical directivity-values for each of the six octave bands 125Hz to 4kHz, see Fig. 4.1. Keywords are case-insensitive, blank lines and comments allowed (introduced by semi-colon). Syntax (keywords in BOLD, values in italic, explanations in plain text, [] items are optional):

162

4. DIRECTIVITY MODULE 15°

horizontal

0°

0°

15°

vertical front

90°

front

right

left

back 180°

270°

90°

top

bottom

Top view

270°

Side view

back 180°

Fig. 4.1 15° source directivity angle convention (SD0).

CATT-SD0 Version Description OctaveBands

=n = loudspeaker name = <[125] [250] [500] [1k]

[2k]

[4k]

[8k]

If electro-acoustic source: Sensitivity = <[S125Hz] [S250Hz] …

[S8kHz] [S16kHz]>

If natural source: NomSPL_at_1m = <[N125Hz] [N250Hz] …

[N8kHz] [N16kHz]>

[16k]>

MaxSPL_at_1m = <[M125Hz] [M250Hz] … [M8kHz] [M16kHz]> Extents = xmin xmax ymin ymax zmin zmax [125 polar data for 125 Hz] [250 polar data for 250 Hz] [500 polar data for 500 Hz] [1k polar data for 1kHz] [2k polar data for 2kHz] [4k polar data for 4kHz] [8k polar data for 8kHz] [16k polar data for 16kHz] CATT-SD0

a file-type tag

Version

a file-format version number, 1 for v7.2, 2 for v8.0

Description

a descriptive name of the source e.g. manufacturer model (max 30 characters).

OctaveBands

defines which of the octave-bands 125 to 16k (version 1 only 125 to 4k) that are measured (in the order of increasing center frequency). The octaves not measured are extrapolated.

163

4. DIRECTIVITY MODULE Sensitivity

source sensitivity i.e. the SPL at 1 m distance on the axis for 1 W electrical input. Only values for the OctaveBands included are specified (in the order of increasing center frequency). Used for electro-acoustic sources instead of NomSPL_at_1m.

NomSPL_at_1m

nominal SPL at 1 m distance on the axis. Only values for the OctaveBands included are specified (in the order of increasing center frequency). Used for natural sources instead of Sensitivity.

MaxSPL_at_1m

max. SPL at 1 m distance on the axis. Only values for the OctaveBands included are specified (in the order of increasing center frequency).

Extents

(for Version >= 1) lists the min-max 3D-extents of a rectangular box surrounding the source. The origin is the acoustic center. The x-axis is forward along the source axis, the y-axis is right to left as viewed from the source, the z-axis is upwards. Note that v8 also allows actual source 3D visuals.

For all OctaveBands polar data for every 15th degree is given.

H125Hz,0° ... H125Hz,180° sym or H125Hz,0° ... H125Hz,345° V125Hz,0° ... V125Hz,180° sym or V125Hz,0° ... V125Hz,345° where Hb,a° horizontal directivity in dB (≥ -15dB, ≤ 50dB) in relation to (0°,0°) for octaveband b and angle a for every 15th degree. Vb,a° vertical directivity in dB (≥ -15dB, ≤ 50dB) in relation to (0°,0°) for octave-band b and angle a for every 15th degree. sym can be entered instead of the remaining values if the directivity is symmetrical The amount of attenuation is in dB relative the on-axis value that should be zero (if the first onaxis value for each band is non-zero the value is subtracted from all sub-sequent values for that band). The resulting numbers must lay in the range -15 to 50 dB. The sign is reversed only not to have to enter a lot of minus-signs.

Hb,180° and Vb,180° must have the same value. Data is specified comma, space or TAB delimited and decimal point must always be used independent of language settings. Sample SD0 TXT-file: CATT-SD0 Version = 2 Description = none OctaveBands = < 125 250 500 1k 2k> ;Note: no 4k 8k or 16k values Sensitivity = < 100.0 100.0 100.0 100.0 100.0> ;electro-acoustic source MaxSPL_at_1m = < 100.0 100.0 100.0 100.0 100.0> Extents = -0.2 0.1 -0.1 0.2 -0.2 0.3 ; >= version 1 125 0 0 0 0 0 0 2.0 4.0 6.0 8.0 9.0 9.0 9.0 9.0 9.0 8.0 6.0 4.0 2.0 0 0 0 0 0 0 0 0 0 0.5 1.5 2.5 4.5 6.5 8.0 9.0 9.0 9.0 9.0 9.0 8.0 6.5 4.5 2.5 1.5 0.5 0 0 0 250 0 0 0 0 0 2.0 4.0 6.0 8.0 10 12.0 12.0 12.0 12.0 12.0 10.0 8.0 6.0 4.0 2.0 0 0 0 0

164

4. DIRECTIVITY MODULE 0 1.0 500 0 0 0 1.0 1k 0 0 0 0.5 2k 0 0 0 1.0

2.0 3.0 4.0 3.0 4.0 6.0 8.0 10.0 12.0 12.0 12.0 sym 0 0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 14.0 14.0 sym 2.5 3.0 5.0 6.0 8.0 8.0 10.0 12.0 14.0 14.0 14.0 sym 0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 16.0 16.0 sym 1.5 3.0 6.0 8.0 10.0 10.0 12.0 14.0 16.0 16.0 16.0 sym 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 18.0 18.0 sym 3.0 5.0 7.0 10.0 11.0 12.0 14.0 16.0 18.0 18.0 18.0 sym

Note that a file must be correct or it will not import. File|Import Directivity|CATT-SD1 TXT

Imports the 10° measured full-space format (*.TXT), see Fig. 4.2. The SD1 text-format is made flexible so that the row order is selectable. Keywords are case-insensitive, blank lines and comments allowed (introduced by ";"). Syntax (keywords in BOLD, values in italic, explanations in plain text, [] items are optional):

10°

top

0° top

rot direction 180°

right

170°

90°

left

front (on-axis) 0°

10°

right

left

arc direction

Back view

bottom 180°

bottom

Fig. 4.2 10° source directivity angle convention (SD1). CATT-SD1 Version Description RotGroups OctaveBands

=n = loudspeaker name = from - to [, from - to ...] = <[125] [250] [500] [1k] [2k]

[4k]

[8k]

If electro-acoustic source: Sensitivity = <[S125Hz] [S250Hz] …

[S8kHz] [S16kHz]>

If natural source: NomSPL_at_1m = <[N125Hz] [N250Hz] …

[N8kHz] [N16kHz]>

MaxSPL_at_1m = <[M125Hz] [M250Hz] … [M8kHz] [M16kHz]> Extents = xmin xmax ymin ymax zmin zmax [125

165

[16k]>

270°


matrix for 125 Hz] [250 matrix for 250 Hz] [500 matrix for 500 Hz] [1k matrix for 1 kHz] [2k matrix for 2 kHz] [4k matrix for 4 kHz] [8k matrix for 8 kHz] [16k matrix for 16 kHz] CATT-SD1

a file-type tag

Version

a file-format version number, 1 for v7.2, 2 for v8.0

Description

a descriptive name of the source e.g. manufacturer model (max 30 characters).

RotGroups

defines how the groups of source rotation data is stored in the file. Angle convention according to Fig. 4.2 is used but an imported file may have another convention. This field describes the convention used in the imported file in terms of the convention used in CATT-Acoustic. Arcs for rotation values not stored in the file are created assuming symmetry. If the data to be imported assumes full symmetry the possible values are: RotGroups = 0-90 RotGroups = 90-0

and 100-180 is created as a vertical mirror copy of 80-0 and then 190-350 is created as a horizontal mirror copy of 170-10. If the data to be imported assumes horizontal symmetry some possible values are: RotGroups RotGroups RotGroups RotGroups RotGroups

= 0-180 = 180-0 = 0-90,180-100 = 90-0,180-100 = 90-0,100-180

and 190-350 is then created as a horizontal mirror copy of 170-10. If the data to be imported assumes no symmetry at all, many combinations are possible, e.g. RotGroups RotGroups RotGroups RotGroups OctaveBands

= 0-350 = 350-0 = 0-180,350-190 = 90-0,180-100,270-190,350-280

defines which of the octave-bands 125 to 16k (version 1 only 125 to 4k) that are measured (in the order of increasing center frequency). The octaves not measured are extrapolated.

166

4. DIRECTIVITY MODULE Sensitivity

source sensitivity i.e. the SPL at 1 m distance on the axis for 1 W electrical input. Only values for the OctaveBands included are specified (in the order of increasing center frequency). Used for electro-acoustic sources instead of NomSPL_at_1m.

NomSPL_at_1m

nominal SPL at 1 m distance on the axis. Only values for the OctaveBands included are specified (in the order of increasing center frequency). Used for natural sources instead of Sensitivity.

MaxSPL_at_1m

Max. SPL at 1 m distance on the axis. Only values for the octave-bands included are specified (in the order of increasing center frequency).

Extents

(for Version >= 1) lists the min-max 3D-extents of a rectangular box surrounding the source. The origin is the acoustic center. The x-axis is forward along the source axis, the y-axis is right to left as viewed from the source, the z-axis is upwards. Note that v8 also allows actual source 3D visuals.

For each octave-band 125 - 16k a matrix of data according to the header definition (RotGroups and OctaveBands) must be included. A file must at least contain values for the range 0 to 90°. Each octave band matrix is preceded by the octave-band label (125, 250, 500, 1k, 2k, 4k, 8k, 16k). The amount of attenuation is in dB relative the on-axis value that should be zero (if the first onaxis value for each band is non-zero the value is subtracted from all sub-sequent values for that band). The resulting numbers must lay in the range -15 to 50 dB. The sign is reversed only not to have to enter a lot of minus-signs. The value at 180° must always be equal for all octave-band arcs. Data is specified comma, space or TAB delimited and decimal point must always be used independent of language settings. Sample SD1 TXT-file: CATT-SD1 Version Description RotGroups OctaveBands Sensitivity MaxSPL_at_1m Extents

= = = = = = =

125 ;according to 0 1.7 4.2 6.3 0 0.8 3.5 5.8 0 0.7 3.2 5.7 0 0.8 3.0 5.3 0 0.6 2.4 4.8 0 0.1 2.0 4.3 0 0.5 1.7 4.0 0 0.1 1.3 3.6 0 0.3 1.5 3.2 0 0.3 1.5 3.0

2 Sample speaker 90-0 ; i.e. data < 125 250 500 1k < 100 100 100 100 < 120 120 120 120 -0.2 0.1 -0.1 0.2

is specified in reverse order from Fig 4.1 > ; i.e. no data for 2k – 16k are available > > -0.2 0.3 ; version 1

RotGroups rows correspond to 8.5 11.7 13.9 16.4 19.6 22.6 8.4 10.9 13.8 16.3 19.1 22.0 8.5 10.5 12.7 15.4 18.5 21.9 7.8 10.0 11.6 14.5 17.8 21.6 6.8 9.2 11.6 13.9 17.4 21.3 6.8 9.6 11.1 13.8 17.6 20.6 6.1 9.0 11.4 14.0 17.5 21.3 5.8 8.0 10.4 13.4 16.5 20.5 5.0 7.2 9.3 12.0 15.2 19.0 4.7 7.5 8.8 11.4 14.4 18.0

90, 80, 70, ..., 0° 26.2 27.9 30.0 31.7 25.3 27.3 29.0 30.7 24.9 27.6 29.7 31.4 24.5 27.1 29.3 30.4 24.2 26.1 27.3 28.1 24.1 25.4 26.5 26.0 24.0 26.4 26.8 26.7 23.5 26.6 28.3 28.3 21.9 25.2 27.7 29.3 21.3 23.8 26.2 27.3

250 ;same principle for 250 Hz

167

rotation 30.0 29.8 30.7 30.9 30.6 31.7 31.7 33.7 30.8 33.2 26.7 28.0 26.2 26.2 27.3 26.0 29.6 30.1 27.5 29.0

28.0 30.3 33.3 33.1 34.2 31.6 29.4 28.5 31.6 34.4

33.6 30.0 31.7 33.6 35.9 34.9 33.2 32.1 31.3 32.0

31.4 31.4 31.4 31.4 31.4 31.4 31.4 31.4 31.4 31.4

4. DIRECTIVITY MODULE ... 500 ;same for 500 Hz ... 1000 ;same for 1kHz ... ;2k-16k ;no data

Note that a file must be correct or it will not import. File|Export Directivity|CATT-SD0 TXT...

Exports the current SD0 or SD2 format to the 15° interpolated text format. If the current format is SD2, the export is made at the selected plot distance (if distance-dependent). File|Export Directivity|CATT-SD1 TXT...

Exports the current SD1 or SD2 format to the 10° measured text format. If the current format is SD2, the export is made at the selected plot distance (if distance-dependent). File|Export Directivity|CATT-SD2 Interpolated...

Exports a current distance-dependent SD2 format to a set of 10° 3D-ballons valid at different distances. A new SD2-file is created based on the CATT_Generic DDI model INTERPOLATED0 that is less accurate than an actually run-time array-modeled SD2 but is much faster and can be used for faster initial aiming. For details see the CATT_Generic help-file. The set of SD1 3D-balloons are placed in CATTDATA-folder SD2DATA\CATT_Generic\Interpolated0

rather than in CATTDATA-folder SD since they do not represent actual sources. These files are named expname_01.SD1, expname_02.SD1 etc. where expname is selected at export. Note: the currently selected Plot distance is used as the far field limit. Utilities|Make Omni

Makes the directivity of the currently selected octave band be omni-directional. Utilities|Copy From

Copies the directivity of another octave-band to the current one. Since most sources have a directivity gradually increasing with frequency, it is often practical to start from the 125 Hz band and then copy and modify the next band until all bands are created. This function is most useful for the SD0 format since it is often entered manually creating generic directivity patterns. Utilities|Calculate DI|Re Axis

Calculates and displays the directivity index, DI, re the axis for the currently selected directivity and octave-band. DI is also shown in prediction output text-files (E_ss_rr.TXT and PARAM_ss.TXT) and with "3D balloons".

168

4. DIRECTIVITY MODULE Utilities|Calculate DI|Re Beam

Calculates and displays the directivity index, DI, re the beam axis (for SD2-sources with beamsteering). Utilities|Copy Name to Clipboard

Copies the currently loaded directivity name into the clipboard. It can then be pasted into CATT-Edit while editing a source-file. Utilities|Protect Data

An option to protect the data in an SD0- or SD1-type directivity file (CLF-format files are always protected). If a file is protected no data can be changed unless it is exported as text and imported again. A protected file shows the protection info in a special field on the dialog. The purpose of the protection is to prevent accidental alteration of files supplied from manufacturers. SD2-files do not need this protection because data is locked in a DLL anyway and can't be tampered with (except for the CATT_Generic DLL that is supposed to be more open). The demo version cannot fully use protected files (CLF-format files are always protected). Utilities|Show Array Elements

Creates a plot-file for showing the location and aim of all array elements as well as their polar patterns and weights. The polar patterns are in |p| rather than dB and 1.0 is at the same scale as 1 m. Any applied weights are indicated by a correspondingly changed radius for the polar pattern concerned. The x-,y-, and z-axis have large tics every meter and small tics every half meter. Lines are drawn from each element location to a point on the array axis located at the selected plot distance from the array origin. The plot-files are named CATTDATA-folder SD\PLT\ELEMENTS_oct.PLT. Note: the element information is not necessarily available for all kinds of DDI array modules. The ARRAY0 model of the CATT_Generic DDI module does however supply the information.

169


4.2 Dialogs SD0, SD1, CF1, CF2 and SD2 formats have several dialog items in common as described after the format specific dialogs.

15-degree interpolated format (SD0) This format requires horizontal and vertical polar measurements for every 15° to be entered or imported.

Symmetrical

Makes the current polar graph be horizontally or vertically symmetrical mirroring the values from the side with the active angle to the other side. The active angle is marked by a black handle. Cylindrical

Creates a cylindrical symmetry by first making the current polar graph symmetrical and then copies the values to the other graph.

170


10-degree measured format (SD1, CF1, CF2) This format requires arcs rotated in 10° steps around the source axis and with values every 10° along each arc to be entered or imported. Note: CF1 and CF2 files are converted internally to SD1 format and cannot be edited, saved or exported. A more extended view is available via the CLFViewer button (assuming the viewer is installed [CLF]). CLF files do not indicate a max SPL per octave band, use the max voltage if available as shown by the CLFViewer.

Slight interface variation for CLF-format files:

Symmetrical

Makes the current directivity be horizontally symmetrical mirroring the arcs on the current side of the sphere to the other side. Cylindrical

Creates a cylindrical symmetry by copying the current arc to all rotation angles.

171

4. DIRECTIVITY MODULE CLFViewer

If a CLF-format file is loaded, the CLFViewer can be launched to give fuller information about the file [CLF]. This function relies on that the viewer has been assigned to handle CF1- and CF2-files as is done by the CLF software installer.

DLL-format (SD2) This format is based on the DLL Directivity Interface (DDI) and requires DLLs that translate or calculate the directivity. DLLs are placed in sub-folders to the CATTDATA-folder SD2DATA. The sub-folders have the same names as the DLLs.

Name (Model)(DLL Directivity Interface (DDI))

The model name, the info field below shows additional model info. Ver (Model)( DLL Directivity Interface (DDI))

The minimum DLL-version that this model's optional data is meant for. Name (Brand)( DLL Directivity Interface (DDI))

The name of the DLL (.DLL removed). Assuming the name BrandX the DLL is CATTDATAfolder SD2DATA\BrandX\BrandX.DLL. All data needed by the particular DLL is placed in its own folder as well as any help available (BrandX.CHM, BrandX.HLP, BrandX.PDF, BrandX.HTM or BrandX.TXT) that is shown when Help is pressed. Ver (Brand)( DLL Directivity Interface (DDI))

Version of the DLL, the info field shows additional DLL info.

172

4. DIRECTIVITY MODULE Help (Brand)( DLL Directivity Interface (DDI))

Displays the DLL help if available. Help in CHM, HLP, PDF, HTM or TXT format (searched in that order) placed in the DLL folder with the same name as the DLL but with .CHM,..HLP,.PDF,.HTM or .TXT extension. This gives a manufacturer an opportunity to include custom made on-line documentation for their loudspeaker models. {

Plot-distance (DLL Directivity Interface (DDI))

For distance dependent models the directivity is plotted and exported at this distance. Accuracy (DLL Directivity Interface (DDI))

For models where accuracy can be selected for a trade-off between speed and accuracy. Far-field at (DLL Directivity Interface (DDI))

For distance dependent models this value marks where the far-field starts (approximately). Ref (DLL Directivity Interface (DDI))

For distance dependent models the reference (what is set to 0 dB when plotting) can be selected. The options are: on axis (1 m)

the reference is the directivity on axis at 1m

on axis (r)

the reference is the directivity on axis at the selected plot distance, r

on beam (r)

the reference is the directivity of the main beam at the selected plot distance, r (same as on axis (r) if the source has no beam-steering indicated). Because the polars are created with 2° resolution a beam angle of e.g. 5° may cause the value at the beam not to be exactly 0 dB.

Note: the air absorption is not applied in polar and balloon plots but only at prediction. (Additional model-dependent data) (DLL Directivity Interface (DDI))

A list-box with the optional model-dependent data along with some edit and manipulation buttons. Press Help to learn about optional data for each model implemented in the current DLL. Update model data

Reads the model-dependent data and passes it to the DLL. Can be used to check the validity of the added or edited data without creating any plot-files.

Elements common to SD0, SD1, CF1, CF2 (and SD2) formats A polar value can be changed either by entering a value in dB in the Value edit-control below the left graph or by direct manipulation of the small handles attached to the graphs. If Enter is pressed, with the cursor in the Value edit-control, the current value is updated and the active angle is incremented. Natural (Source type)

A natural source such as a talker, singer or an instrument.

173

4. DIRECTIVITY MODULE Electro-acoustical (Source type)

An electro-acoustical source such as a loudspeaker or loudspeaker array. Sensitivity/Nominal SPL 1m (Source type)

The sensitivity information is not directly used except in the but source-file using Lp1m_a = Lp_Sensitivity but should be given for reference and future use, for natural sources the nominal (typical) spectrum can be used in the source-file using Lp1m_a = Lp_nominal. Notes: CF1 and CF2 files of type "powered" or "active" do not have Sensitivity. For SD2 this data cannot be changed via the dialog displayed since it is read from the DLL. Max. SPL (Source type)

Max SPL at 1m on axis. These values are compared to the Lp1m values entered in the source-file. These maximum level can be set in the source-file using Lp_max. None (Visuals)

When plotted, the source has no visuals except for a point and the source id. 3D Extents... (Visuals)

For sources the 3D extents (a min-max box) can be set (relative the acoustic center). This will serve as a reference and give a feeling for the actual size of the source in plot-files as well as VRML exports and SHADED.OGL. Note that for SD2 the extents can not be changed from the dialog like SD0 and SD1 since the extents of an SD2 is typically taken from the DLL (for CATT_Generic models, the extents may be entered in the Additional modeldependent data list but typically it will be given by the DLL in question and cannot be altered). 3D marker (Visuals)

The source visual is defined by the selected CATT marker-file located in the CATTDATAfolder SD or a sub-folder of SD. Marker-files can be created in GEO-format using the NOROOM directive and be exported using Prediction module File|Export|CATT Marker MRK. The coordinate system used should be x as the source axis, z up and y to the left and the origin at the acoustical center, see dialog above for 3D extents. Note: For SD2-files the marker-files (if any) are supplied by the DDI module DLL. CLF files may have 3D cabinet data embedded and it is then automatically converted to an MRK-file the first time the Visuals button is clicked (CLF files with no cabinet get a +/-1 dm box).

174

4. DIRECTIVITY MODULE Description

For reference purpose e.g. for the real name of the loudspeaker or source e.g. "Female talker" or "Horn, type 15" (SD0/SD1/SD2). (Octave-band)

The radio-buttons selects what band to edit (SD0/SD1) and/or to create 3D balloons for. If bands are extrapolated or estimated it is indicated. Locked

When the check-box is selected no change of directivity data can be made but only inspection. Each time a new directivity is loaded or imported, editing is initially locked to prevent accidental changes. See also Utilities|Protect Data. For plot-files this module will create a CATTDATA-folder SD\PLT where all plot-files are created. (Plot)

Note: for SD2 arrays with a distance dependent directivity, the air absorption is not included in polars and balloons even if Plot distance is set far away but is added only at prediction. Oct (Balloon) (Plot)

Creates plot-files for the interpolated 3D directivity for the selected octave-band (SD\PLT\BALLOON_oct.PLT). The directivity is sampled at 7.5° steps for SD0 and 5° for SD1, CF1 and CF2 while SD2 has an option of 5° or 2° (Preferences) and is thus not simply a plot of the entered values. For SD0 seamless interpolation is inherent in the procedure employed, for SD1 3D interpolation is performed between the closest 4 data points and for SD2 (depending on the DLL, this may either be exactly simulated angles or be interpolated from measured data). A Preferences option allows for balloons to be colors-shaded. All (Balloon) (Plot)

Similar to Oct above but all octave-bands are plotted, each as sub-frame in a single file. The file-name is BALLOONS.PLT. Polars (Plot)

Creates a plot-file for all octaves (SD\PLT\POLARS.PLT) with polar directivity plot-file octave bands 125Hz - 16kHz. For SD0 and SD1 the data at is plotted from the data set at the respective resolution while SD2 is plotted at 2° resolution. Contours (Plot)

Creates a plot (CATTDATA-folder SD\PLT\CONTOURS.PLT) of the -3, -6 and -9 dB contours at selected distance and angle to a fictive audience plane. For details on directivity contours, see Section 4.3.

175


Visuals (Plot)

Creates a plot-file (SD\PLT\VISUALS.PLT) for the source visuals (if not None).

4.3 Theory Five types of directivities are available. Type SD0 is interpolated from horizontal and vertical polar measurements entered at 15° steps. This format is most suitable for sources that are entered manually from polar data supplied on loudspeaker data-sheets or for creating generic loudspeaker types. Type SD1 is based on full space measurements in 10° steps (often assuming symmetry so that only a fourth or half of the values are actually measured), intermediate values are interpolated. Type CF1 and CF2 [CLF] are measured at 1/1octave:10° steps and 1/3-oct:5° respectively and are converted internally to SD1 format. Type SD2 is based on the DLL Directivity Interface (DDI) defined by CATT and is capable of arbitrary data-format translation, array modeling and other special functions.

The 15-degree interpolated format (SD0) The directivity approximation can handle any type of polar patterns, also non-symmetrical ones. The only constraints are to ensure consistent and logical patterns: 1) the on-axis value must be 0 since it is the reference (enforced); 2) the horizontal and the vertical value at 180° (in the back) must be equal (enforced); 3) values must be in the range -50 dB to 15 dB .

The 10-degree measured format (SD1) Can handle any type of polar pattern, also non-symmetrical ones. The only constraints are to ensure consistent and logical patterns: 1) the on-axis value must be 0 since it is the reference (enforced); 2) all values at 180° (in the back) must be equal (enforced);

176


3) values must be in the range -50 dB to 15 dB.

The any-degree measured, interpolated or simulated format (SD2) The SD2 directivity format is based on a DLL interface rather than a data format. This means that the data that a DLL interfaces with may: 1) be measured, analytical or simulated at any resolution; 2) be based on unorthodox spacing (e.g. 1° in the front and 10° in the back of a directive speaker); 3) have > 15 dB overshoot. 4) be distance dependent e.g. for arrays that typically do not have 1/r² behaviour and that may have very long near-fields (up to 100 m is not uncommon) 5) be in higher frequency resolution that 1/1-octaves (v8 still only handles 1/1-octaves but the SD2-format opens up for a future higher resolution) To achieve the above it is "only" a question of writing a DLL for it since the interface towards the prediction software is the same. Prediction using array-modelling takes at least one magnitude longer with a full array modelled (say 16 transducers that all call an internal DLL SD0-, SD1- or CLF-version of the transducer directivity and for every 1/6-octave it makes a summation for the octave band value so it has to take some extra time). On the positive side is that when predicting using high directivity fewer rays can be used since the direct sound (that is not solved by rays) is dominating and that compensates some for the longer calculation.

Note regarding auralization with calibrated SD2 sources Some SD2 format sources are absolute calibrated (typically arrays, see each individual DDI module help-file) and use the actual assumed Lp1m_a in the source-file that should look like this: srcid x_pos y_pos z_pos model.SD2 x_aim y_aim z_aim or aim(H,V) Lp1m_a = Lp_voice_normal or other desired acoustic input Gain_a = < Gain_a-values <= 0 > use negative values to attenuate below max Delay_e = delay

With Gain_a all zeros, the predicted SPL will correspond to the max possible SPL with pink input no matter the values of Lp1m_a. Gain_a can be used with negative values to control how much below max to operate. The reason for giving an actual Lp1m_a spectrum that does not affect the SPL is because this spectrum must be removed for auralization since it is later to be replaced by the spectrum of the anechoic file at convolution. It also makes it possible to mix calibrated and non-calibrated sources in auralization (e.g. a main array and delayed under-balcony fill speakers) as long as they all use the same Lp1m_a. In prediction result files, the source level at 1 m is not indicated for a calibrated SD2, instead the Gain_a values are shown.

177


Directivity Index The Directivity Index (DI) is: DI = 10 log10(Γ) where Γ is the directivity factor defined as in [Beranek], p109: ...the ratio of the intensity on a designated axis of a sound radiator at a stated distance r to the intensity that would be produced at the same position by a point source if it were radiating the same total acoustic power as the radiator. Free space is assumed for the measurements. Usually the designated axis is taken as the axis of maximum radiation, in which cases Γ always exceeds unity...

DI is always calculated on the main axis and if directivities are entered with the maximum radiation off axis, DI may become negative. This causes no problems in the predictions. For SD2-sources that have beam-steering, the DI re the beam angles can be displayed. To obtain the total acoustical power and DI of a source and octave-band the directivity is sampled at 412 uniformly spaced points for SD0 and SD1 (4π/412 steradians) and at 1652 points for SD2 (4π/1652 steradians).

Directivity Contours Directivity contours are created by backing off from the axis until -3 dB (for the -3 dB contour) is reached. This is done for every 5° around the axis and by stepping back 1° at the time. Since the contour calculation starts out from the axis, sources that do not have their maximum radiation on-axis will have somewhat misleading contours, see Fig 4.1. for a –3dB contour example.

Fig. 4.1 Contours for max radiation off-axis

Caching of DI and contours The calculation of DI and contours can be time-consuming for some types of sources so the results are cached in a parallel file named after the directivity file, e.g. for a myarray.SD2 file the cached data will be stored in myarray_SD2.SDC. The actual caching takes place the first time a directivity is used in prediction and is updated if the directivity file has changed. SDC files can be deleted at any time and will be created when needed.

178

5. PLOT-FILE VIEW/PRINT MODULE

5. PLOT-FILE VIEW/PRINT MODULE CATT-Acoustic creates four types of plot-files. Static 2D, static 3D and 2D-, 3D-files with dynamic sub-frames. Depending on the type, different buttons on the optional Plot-file control dialog and the toolbar are enabled. The plot-file type is displayed in the left status bar after the plot-file name.

The plot-file viewer module has a near identical standalone version, CATTPLT32.EXE, that can be loaded from the File|External CATT Tool menu and can be used e.g. for viewing and comparing plot-files side by side. If a plot-file called startup.plt is found in the CATTDATA-folder it will be shown at startup. When the mouse is passed over 2D graphs, maps and geometries the left status-bar shows values corresponding to the current mouse position:

•

2D geometry projections shows ( x ; y ; _ ), ( x ; _ ; z ) or ( _ ; y ; z ) depending on view,

•

graphs show (x_value ; y_value),

•

2D maps show (x ; y ; map_z ; map_value).

In addition to viewing, printing, and exporting plot-files, the module can also play Windows WAV-files. A Preferences option allows the View module to be automatically minimized when another module becomes active or to not show the Plot-file control.

5.1 Menus File|Open Plot-file...

Opens a particular plot-file (.PLT) for viewing. File|Save Plot-file

Saves the current plot-file (only meaningful for 3D plot-files - the perspective will be saved). File|Save Plot-file As...

Save current plot-file under a new name. File|Export Plot-file To|Meta...

Exports the current plot-file to the 32-bit Enhanced Windows Meta-file format to insert in e.g. MS Word. Displays a dialog for Meta-file settings. File extension is set to .EMF. See also Edit|Copy below.

Plot width

Windows Enhanced Meta-file export

Width of the plot-area to use, in mm. The height will be calculated to preserve true aspect.

179

5. PLOT-FILE VIEW/PRINT MODULE File|Export Plot-file To|Bitmap...

Exports the current plot-file to Bitmap (BMP) file for import into e.g. MS Word. Depending on the type of plot-file a 4, 8 or 32-bit color file is created. File|Export Maps For Spreadsheet...

If the current plot-file contains one or more audience area maps they can be exported to a semi-colon delimited text-file for import into e.g. MS Excel, the maps will be written including x, y coordinates and some general info. If the complete data set is marked and a “Surface graph” is chosen it can be viewed inside Excel. Map points with no data will show up as blank cells in Excel. This is a function requested by several users but using Excel's “Surface graph“ instead of the actual map in a PLT-file is not recommended, see below. There may be other reasons to export the data though. Note: The map values are upside down in Excel as compared to the map shown by the PLTviewer. This is because a different convention in Excel but a “Surface graph” will have the same principal look as the original in the PLT (since the x-y coordinates are given they will help the interpretation). However, carefully note that the Excel “Surface graph” interpolates the values to create a nice-looking smooth graph and that can sometimes be deceptive. For maps with a small Step value the x- and y-values on the graph axis must be set as “Numbers” with say 2 decimals to display well. A further note is that the actual data values are located at the crossing of the lines in the Surface graph rather than in the middle of each map square as in the PLT-file. File|Create Header Plot-file...

Creates a plot-file with only a header text displayed in the center. This can be used e.g. to create a plot-file list interleaved with short descriptions. If a list-file is currently open the header file is inserted in the list after the currently selected plot-file (to simplify creation of lists with header plots). To edit the list, see List menus below. File|Page Setup... File|Print Plot-file...

Prints the current plot-file to the printer. The print-out is WYSIWYG so if you zoom a 2D plot also the print-out will be zoomed. To change printer, paper, size, or orientation, use Page Setup that displays a standard Page Setup dialog for selection of printer, page size, margins and orientation. These settings are local to the software and are kept until the application is terminated. Since the plots have a fixed aspect ratio (640x385) the print-out is maximized and placed towards the top margin of the selected print area.

margins

plot

plot plot

plot

Portrait mode

margin

title

paper

Landscape mode

Edit|Copy As 32-bit Meta

Copies the current plot-file into the Windows Clipboard in Enhanced (32-bit) Meta-file format.

180


The picture can then be pasted into some 32-bit programs (e.g. MS Word or PageMaker). The size of the plot is preserved. Note: the size of the plot-file viewer window is displayed in the right status bar. This makes it easier to copy/paste using a particular size since the size can be noted and if a new export/copy must be done with the same size, the window can be resized to match the previous export/copy. Edit|Copy As Bitmap

Copies the plot-window to a clipboard bitmap. Palette information is included so that true colors are preserved. The size of the plot is preserved. See note above about plot size. Edit|Copy 3D-view to PRD

Copies the current 3D view parameters to be the default settings in the current PRD-file (the Geometry view/check dialog) List|Open List-file

Opens a text-file with a list of plot-files (and optionally WAV-file) names. Each time plot-files are created a list of all the names are written to a module-dependent text-file (e.g. PLT_GEO_FULL.TXT for the Prediction module). The View module then receives a message to display the list. The format of a list-file is:

[PATH]PLOTFILE.PLT ...

[[,][PATH]WAVFILE.WAV]

where the WAV-file is optional. If folder- or file-names contain spaces the whole path and name must be enclosed in "". Sample list-file: C:\temp\view4.PLT C:\temp\shaded.PLT c:\temp\sound.WAV "C:\temp\view4 old.PLT" "C:\temp\shaded old.PLT" "C:\temp\sound old.WAV" "C:\temp\shaded old.PLT" C:\temp\sound.WAV "C:\temp\shaded old.PLT", C:\temp\sound.WAV

Note: if edited manually, be sure to include the full path to the files so that the viewer always can find them, if no path is given the viewer assumes the files to be in the same folder as the list-file. List|Save List-file

Saves the current list-file. List|Save List-file As

Saves the current list-file under a new name. The current plot-file then becomes the first file in the new list. List|Insert Plot-file...

Inserts a new plot-file after the current one. To insert multiple files use Ctrl-LMB or Shift-LMB.

181


List|Remove Plot-file

Removes the current plot-file from the list. Note: if only one file remains in the list it can not be removed. List|Associate WAV-file

Associates a WAV-file with the plot-file. The WAV-file can be played by pressing the Play WAV-file button on the Plot-file control dialog or the toolbar. When Options|Auto Play WAV is on the WAV-file is played automatically when the plot-file has loaded. List|Deassociate WAV-file

De-associates the WAV-file associated with the current plot-file. List|Prev Plot-file, List|Next Plot-file

Display previous/next plot-file in the list (PgUp/PgDn). The File << and >> buttons on the Plotfile control dialog and the toolbar duplicates these actions. If the previous/next plot-file has an associated WAV-file and Options|Auto Play WAV-file is on the WAV-file is played. List|Select Plot-file...

Displays a small list-box where plot-files in the list can be selected and the list be edited. List|Auto View List

Toggles auto view the list of files the list (Ctrl+A). Use Options|Auto View Delay to set interfile delay. If plot-files have associated WAV-files and Options|Auto Play WAV-file is on WAV-files are played automatically. List|Print All Files In List

Prints all plot-files in the list with an option to skip files. List|Export All Files In List To|

Exports all plot-files in the list to either Meta or Bitmap with an option to skip files. List| Export All Maps In List

Has the same functionality as Print All Files In List but each plot-file in the list that has an exportable map will instead be exported with the extension changed to .TXT. In other ways similar to File|Export Maps For Spreadsheet. Options|Plot-file Controls

Show/hide the floating dialog (Ctrl+R). Most functions on the control can be accessed via the toolbar or by direct mouse-actions on the plot window. A Preferences option can be selected so that the dialog is never opened. Options|Auto Play WAV-file

Automatic play of a WAV-file attached to a plot-file when it is loaded.

182

5. PLOT-FILE VIEW/PRINT MODULE Options|Auto View Delay...

Selects the inter-file delay (in ms) for plot-file list auto view. Options|Select Frame...

Displays a small list-box where any plot-file frame can be selected. For 3DA files with shaded polygon frames (notably PLINFO.PLT) can be selected by the left mouse button by pressing “F” for Frame, pressing “F” again returns to normal mode.

Options|Map Scale...

Choice of scale for color mapping. Use original scale in plot-file

Mapping scales

If this option is checked the original scales selected in Audience area mapping dialog is used instead of the ones selected here. Options|Map Smoothing

2D maps can optionally be smoothed to an apparent double resolution. The procedure chosen does not actually paint a very false and too beautified picture but is rather just easier on the eye. The procedure is as follows:

⋅

each map square is divided into 9 parts:

183


1/4 1/2 1/4 +---+------+---+ | | | | +---+------+---+---+ | | A | B | C | | |as is | | | +---+------+---+---+ | | | D | E | +---+------+---+---+ | F | G | +---+---+

B' => (B+C)/2

D' => (D+E+F+G)/4

⋅

the center square (A) is left as is

·

each side rectangle (B) is mixed with its neighbouring rectangle (C)

·

each corner square (D) is mixed with its three neighbouring squares (E,F,G)

Note: 3D rotatable maps are not smoothed. Options|Map Palette|...

Choice of palette for color mapping. The status line at the bottom of the frame window spells out the abbreviations. Options|Shaded 3D Palette|...

Choice of palette for shaded 3D plots. The status line at the bottom of the frame window spells out the abbreviations. Options|Remove 3D Shading

Converts a shaded or colored 3D plot to wire-frame. The process is irreversible but the file on disk is not changed. To permanently convert a file to wire-frame it must be saved after conversion.

5.2 Dialogs Plot-file control Most functions can also be performed with click-drag with the mouse directly on the plot window where the value under mousecontrol is updated according to the speed of mouse movement. Angle and movement step factor is according to the value selected on the plot-file control or the toolbar. In the descriptions below LMB means Left Mouse Button and RMB

184


General buttons << File >>

Display previous/next plot-file in a list (duplicates the function of menuselections List|Prev Plot-file / List|Next Plot-file). (toolbar)

Reset

Reset to initial or saved settings

Play WAV-file Plays a WAV-file if attached to a plot-file. (toolbar)

2D buttons (Zoom)

Zoom out/in. (toolbar)

Scroll bars

Moves the plot

Scroll arrows Moves the plot Click in the plot window. When the plot is currently not zoomed: zooms and moves directly to the corresponding fourth of the screen; when the plot is currently zoomed: the zoom will be reset (same as the Reset button); If the dialog is closed, or not activated, pressing the “1”, “2”, “3” or “4” on the keyboard while the plot window is active gives the same result. If 2D plots are zoomed, pressing/releasing the SHIFT key toggles the size of the text.

2DA buttons 2D buttons plus: < Frame >

Previous/next "frame" in the animation list. If the dialog is closed, or not activated, left and right arrow keys performs the same function allowing for basic list operation with PgUp/PgDn and arrow keys. (toolbar). If the mouse has a wheel it can also be used to change frame as can LMB dragging across the yellow title bar. Using Options|Select Frame any frame can be selected

3D buttons When a button is pressed, the current value is displayed. (Dist.)

"Distance", move closer/farther (for Camera and Camera Set move forward/ backward). (RMB + toolbar)

(Lens)

Lens angle smaller/larger for Natural, Camera and Camera Set (Ctrl-RMB + toolbar).

(Move)

Horizontal arrows: moves the picture left/right (for Camera and Camera Set moves the camera left/right). Vertical arrows: moves the picture down/up (for Camera and Camera Set moves the camera down/up). (Ctrl-LMB + toolbar)

(Rotate)

Horizontal arrows: horizontal view angle (for Camera and Camera Set rotate camera horizontally). Vertical arrows: vertical view angle (for Camera and Camera Set rotate camera vertically). (LMB + toolbar)

185

5. PLOT-FILE VIEW/PRINT MODULE (Step)

Decrease/increase factor when selecting e.g. Dist. (1/64 to 64). If the dialog is closed, or not activated pressing keyboard < decreases and pressing > increases the step (can be done while click-dragging) (toolbar).

(View) Natural

Selects Natural point perspective. If the dialog is closed/not activated press “N”.

Parallel

Selects Parallel perspective. If the dialog is closed/not activated press “P”. When in parallel perspective, pressing “I” selects an isometric view.

Camera

Selects Camera perspective. If the dialog is closed/not activated press “C”.

Set

Shows and moves the camera in the current perspective (Natural or Parallel). If the dialog is closed/not activated press “S”.

“X”, “Y”, “Z” Pressing these keys align the perspective along the X, Y or Z axis, most useful with a parallel projection e.g. when viewing directivity balloons.

3DA buttons 3D buttons plus: < Frame > See 2DA buttons above. “F”

Pressing “F” will enable by the LMB selection of frames with shaded polygons (notably PLINFO.PLT), pressing “F” again returns to normal mode. If the plane has sub-divisions they follow directly after the plane-frame and can be moved to as usual e.g. by the mouse wheel or the right arrow key. While in Frame selection mode - "select frame" is append to the text in the left status bar.

The extensive toolbar duplicates most controls on the floating Plot-file control dialog:

The toolbar tool-tips display the current value for a 3D setting (the current camera position and target for camera mode, i.e. the aim line end-points as seen in the Camera Set mode)

186

6. SOURCE ADDITION MODULE

6. SOURCE ADDITION MODULE The Source addition module is used for adding echograms from Full detailed calculation (or for special cases from Early part detail ISM) for multiple sources when analyzing one or more PA-loudspeakers interacting with the natural speaker and similar situations. The module can optionally create ECH-files for the Post-processing module so that multiple source auralization can be fine-tuned without requiring a new complete calculation for every source change. Note: The actual addition for multiple source auralization is made after Post-processing by adding created SIM- or WAV-files. Several source parameters can be varied without the need for a recalculation in the Prediction module. The files created by the Source addition module follow basically the same principles for name syntax as for the Prediction module, see 6.3. Note: for purely numerical estimates the Audience area mapping directly predicts maps for multiple sources with delays. To use this module the (Save data for) setting in the Prediction dialog for Full detailed calculation must have been selected at prediction (for special cases also the Early part detailed ISM can be used). The Prediction module then creates files with the name syntax E_ss_rr.ADD which are read by the Source addition module. With the exception of source directivity-files, ADD-files contain all necessary information to perform the source addition. When prediction is made with (Save data for) Source addition selected, a CMB-file is generated for the calculation case with the same base-name as the current source-file (typical name SRC.CMB). A CMB-file can also be generated with only a geo-check by selecting Create .CMB in the Geometry view/check dialog. This can be used to change the source aims and, at the same time, generate SRCINFO plot-files where all source aims can be seen. The module always works with files in the directory where the CMB-file was loaded from and when a CMB-file is first loaded all ADD-files are searched and the module figures out which source-receiver combinations are possible. Also the geometry check-sums are compared. No information regarding source data (other than source locations) are saved in the ADD-files. It is possible to create each source separately and then later add them together as long as the geometry check-sums and the head-direction match and ADD-files have been created. The quality of the calculations is up to the user to decide as no comparison is made concerning number of rays used etc. The following parameters can be varied for each run: ·

source combination (i.e. which sources to add together)

·

source directivities

·

source aims and roll angles

·

source level spectrum

·

electrical delays

·

background noise level (for RASTI and STI).

·

a project name

·

receiver combination (i.e. which receivers to use)

·

summation of direct sounds with or without phase.

Zero time (from where D-50 etc. is calculated) is set to when the first sound arrives (electrical

187


delay taken into account) i.e. the "law of the first wavefront" is used. The length of the sum echogram created is set from the shortest of the added echograms (delay taken into account).

6.1 Menus File|Open Combination...

Opens an existing source combination/data CMB-file. File|Save Combination

Saves the current source combination/data CMB-file. File|Save Combination As...

Saves the current source combination/data under a new name. Utilities|Latest Results List

Each time text-based result-files are created the corresponding filenames appear in a small list in the main window. This item toggles the list on and off while preserving its contents. To load a file for viewing in CATT-Edit double-click the filename. Utilities|Show Source Combination...

Displays an overview over currently selected source combination settings. Example: Project: Example --------------------------------------------------------------------------------------------------S# Directivity Aim_x Aim_y Aim_z 125 250 500 1k 2k 4k 8k 16k Delay ................................................................................................... A0 LSP1.SD0 1.0 3.7 1.7 70.0 73.0 76.0 79.0 82.0 95.0 95.0 95.0 0.0 A1 LSP2.SD0 1.0 3.7 1.7 70.0 73.0 76.0 79.0 82.0 95.0 95.0 95.0 0.0 A2 LSP3.SD1 1.0 3.7 1.7 70.0 73.0 76.0 79.0 82.0 95.0 95.0 95.0 0.0 ................................................................................................... S# Hor Ver Roll ...................... A0 -0.0 0.0 0.0 A1 -26.6 0.0 0.0 A2 45.0 0.0 0.0 ---------------------------------------------------------------------------------------------------

This is then followed by delay matrices (delays for all source-receiver combinations, one for the initial delay only and one for initial delay plus electrical delay). The source aim angle definition (Hor, Ver, Roll) is documented in section 2.4 about SRCINFO.PLT. Utilities|Addition-file Info...

Displays essential information from an ADD-file created by the Prediction module and that is read by the Source addition module (reverberation time, etc.). Example:

188

6. SOURCE ADDITION MODULE Essential ADD-file info -------------------------------------------------------------------------------------Source location [m] : ( 1.000 1.700 1.700 ) Receiver location [m] : ( -3.000 20.000 1.300 ) Early part trunc. [ms] : 178 Early part max order : 18 No of reflections : 17111 (all 8 octaves) + direct Mean free path [m] : 7.49 Mean abs. coeff. [%] : < 18.77 18.37 19.68 20.02 19.69 20.15 20.96 24.57 > Reverberation time [s] : < 1.44 1.47 1.35 1.30 1.27 1.09 0.72 0.33 > Background noise [dB] : < 45.00 38.00 32.00 28.00 25.00 23.00 21.00 19.00 > -------------------------------------------------------------------------------------

6.2 Dialogs The main Source addition dialog controls the combination of sources and receivers to add echograms for.

(Sources)

Enable/disable the corresponding source(s). Show all (Sources)

Displays all currently selected and enabled sources. Add direct sound with phase (Sources)

With this option selected direct sounds from each of the sound sources are allowed to interfere (two equal in-phase source contributions give close to +6dB while out-of-phase signals can be -10 dB down or more). Without this option the direct sound is, just like higher order reflections, added on a power basis (two equal source contributions always give +3dB). The interference calculation is based on digital octave-band filters and not only on the octave-band center frequencies (i.e. the interference is calculated with full bandwidth but the result is filtered in octave-bands so interference patterns will be somewhat smoothed). With this option, the direct sound in the echograms displayed will appear to have a short decay due to the digital filters used but it is numerically insignificant. Note: the addition with phase affects LF

189


values too since a figure-of-8 microphone is simulated for the lateral sound taking into account the reversed phase of reflections registered at the left side of the microphone. LFC is not calculated if direct sound is added with phase. (Sources)

Displays a dialog where the source data is selected: SPL at 1m distance, eq/gain, electrical delay, source directivity, source aim coordinate and source roll angle.

(Level control, dB)

Data source

For a natural source only Lp at 1m acoustic (Lp1m_a) can be varied, for an electro-acoustic source all three (Lp1m_a, Gain_a, Lp1m_ea) can be varied and a change in one causes a change in one of the other according to: Change in Gain_a → Change in Lp1m_a → Change in Lp1m_ea →

Lp1m_ea = Lp1m_a + Gain_a Lp1m_ea = Lp1m_a + Gain_a Gain_a = Lp1m_ea - Lp1m_a

Clicking the scrollbar arrows changes a level with 0.2 dB. Directivity

Data source

To select OMNI natural directivity select (none) after using the Any format… button. Aim Point / Angles

Data source

As an alternative to specifying the aim point, aim angles can be used. The angle conventions are the same as for the aim() function in the source file. When switching aiming mode from point to angles, the aim vector length is calculated and saved so when switching back again the new aim point is at the same distance from the source as before (just switching aim mode changes nothing).

190

6. SOURCE ADDITION MODULE Receivers

Selection of the set of available receiver positions to use. Show all

Receivers

Displays all currently selected and enabled receivers. Different types of output-files are created depending on if a single receiver is selected or if multiple receivers are selected, see Section 6.3. Note: all parameters are calculated with the direct sound from the source that arrives first as the reference (utilizing the "law of the first wave-front"). Show source aims

Receivers

Creates a simplified plot-file over the geometry showing a wire-frame room view and the chosen source aims. The plot-file created for this purpose will have the same base name as the current CMB-file. Additional plot-files are created automatically, see Section 6.3. For a more detailed view of the how the sources are aimed, create a CMB-file together with SRCINFO plot-files using the Prediction module. (Octave-bands (for plots))

Selection of octave-bands for detailed output. Text-file results are created for all bands. Project

A comment describing the project. Create E_ss_rr.ECH

Creates files for the Post-processing module. In the files, information on source eq. and delay is contained which can be taken into account at post-processing using the SIM- or WAV-file addition utilities (Post-processing:Utilities|SIM-file addition) or WAV-file addition utility (Postprocessing:Utilities|WAV-file addition). These ECH-files are identical to those that may have been created by the Prediction module at the same time as the ADD-files were created, except for any source changes applied in the addition module. Bkg noise change

This option enables a change of background noise level to study the effects on RASTI and STI. This option or the values given are not saved in the CMB-file but are just to be used for experimenting. If the option is off, the background levels as found in the ADD-files from prediction are used. Note: if individual receiver background noise has been used it will apply also with Source addition until Bkg noise change is selected where all receivers will get the selected noise.

191


6.3 Output-files This section contains a list of all result plot- and text-files created by the Source addition module. It complements the general overview of file-types in Section 1.4. For the sample output text-files below, the case in Fig. 6.1 has been used.

Z

Z

Y

X

5m Audience:240m²

Volume:1921m³ (approx.)

01 B1

Y A0 02

X

Fig. 6.1 Sample source addition configuration for two sources (A0 and B1) and two receivers (01 and 02).

General MULTPLT.TXT

list of the latest result plot-files.

E_ss_rr.ECH

optional echogram files for the Post-processing module.

src.PLT

source aim view with simplified geometry here src is taken from the name of the current CMB-file.

Single receiver result-files If a single receiver is selected detailed data for that receiver is created in two types of result files.

E_nn_rr_oct.PLT (nn is 01, 02, 03 etc. depending on the number of sources used and with 8 sources per page). In each small diagram both the echogram of the specific source (solid lines) and the total echogram from all used sources (dashed lines) can be seen. In each diagram the direct sound from the corresponding source is marked by a small circle. The responses are integrated using a time-constant roughly corresponding to the length of the response divided by 500 (for typical 300 ms responses the time constant

192


will be 0.6 ms).

E_rr_INT.PLT

gives multiple source decay curves T-30 and EDT estimates. This decay may be useful with large systems and long geometrical delays. When multiple sources are used (especially widely spaced) the RT is no longer only a room property but a property of the electro-acoustical system as a whole. Note that the latest part of the decay may have a steep slope that is caused by limited lengths of individual originally calculated full echograms. If the late part data is missing (e.g. when using Early detailed ISM for special purposes, no decay file is created). Note: EDT is given but a multiple source decay can be very uneven in the early part and then the EDT value be of little use since according to the ISO standard it is calculated using a straight line regression and if the decay is not very straight EDT can have an unexpected value.

E_rr_EKgrad.PLT EKgrad calculated from the sum echogram (note that any values > 2.0 are plotted at 2.1 to be able to keep a fixed scale). The EKgrad for the source sum echogram potentially indicating problems with echoes due to inappropriate source delays, levels or aiming. No values are given in the E_rr.TXT result file since it is then better to look at the graphs. E_rr.TXT

contains source combination/data and results in the form of acoustic parameters. The parameters included are D-50, C-80, LFC, LF, Ts, SPL, RASTI and STI with and without background noise. 8k and 16k Hz values can optionally be included (see Preferences). Example:

CATT-Acoustic v8.0a : source add. acoustic param., Copyright © CATT 1988-2007 =================================================================================================== Original file name: C:\CATTDOCS\v8\SHOEBOX_multsrc\OUT\E_01.TXT Original save time: 2007-03-25 12:04:21 --------------------------------------------------------------------------------------------------Project : v8 Shoebox Receiver id : 01 Location : -3.00 20.00 1.30 Head direction : -3.00 19.00 1.30 SOURCE COMBINATION --------------------------------------------------------------------------------------------------S# Directivity Aim_x Aim_y Aim_z 125 250 500 1k 2k 4k 8k 16k Delay ................................................................................................... A0 CATT.SD0 1.0 4.7 1.7 70.0 73.0 76.0 79.0 82.0 95.0 95.0 95.0 0.0 B1 CATT.SD1 -2.0 7.7 1.7 70.0 73.0 76.0 79.0 82.0 95.0 95.0 95.0 0.0 S# Hor Ver Roll ...................... A0 -0.0 0.0 0.0 B1 -0.0 0.0 0.0 Direct sound power addition. DIRECT SOUND DELAY/LEVEL (re first arrival) --------------------------------------------------------------------------------------------------S# delay 125 250 500 1k 2k 4k 8k 16k sum ................................................................................................... A0 9.5 -3.3 -1.1 -1.3 -4.3 -0.1 -0.8 -1.6 -3.1 -1.4 B1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Sound from Src B1 arrives first (becomes the reference). Sound from Src A0 arrives 9.5 ms after Src B1 and with relative level 125 Hz to 16 kHz <-3.3 -1.1 -1.3 -4.3 -0.1 -0.8 : -1.6 3.1> dB i.e. slightly weaker than that of Src B1.

193

6. SOURCE ADDITION MODULE PARAMETERS --------------------------------------------------------------------------------------------------Parameter 125 250 500 1k 2k 4k 8k 16k sum ................................................................................................... Ts [ms] 90.4 79.3 71.9 64.2 56.2 53.9 40.7 28.9 45.4 D-50 [%] 42.4 49.3 54.5 59.5 69.9 70.8 75.7 90.8 75.3 C-80 [dB] 2.5 3.5 4.4 5.5 6.3 6.4 10.2 16.6 8.5 LFC [%] 39.5 39.7 37.3 30.5 29.0 31.5 37.9 32.2 34.5 LF [%] 23.8 24.1 22.4 17.5 15.1 16.6 20.3 15.1 18.0 SPL [dB] 61.9 62.7 63.7 64.3 65.8 79.2 79.5 75.5 83.4 82.8 (A) --------------------------------------------------------------------------------------------------STI orig [%] without / with bkg. noise : < 45.0 38.0 32.0 28.0 25.0 23.0 : 21.0 - > dB ................................................................................................... 63.5 / 63.3 --------------------------------------------------------------------------------------------------RASTI [%] without / with bkg. noise : < - - 32.0 - 25.0 - > dB ................................................................................................... 62.5 / 62.5 --------------------------------------------------------------------------------------------------125 250 500 1k 2k 4k 8k ................................................................................................... Weights 0.13 0.14 0.11 0.12 0.19 0.17 0.14 (original) ---------------------------------------------------------------------------------------------------

Parameters are calculated assuming the sound from the source arriving first as the reference. SPL is the total SPL of all sources.

E_rrX.TXT

same as E_rr.TXT but made for MS Excel import (text is within "" and data is semi-colon delimited).

Multiple receiver result-files If multiple receivers are selected two types of overview result files are created. TREND1,2.PLT

this file is similar to TRENDn_ss.PLT created by the Prediction module, as it concentrates the results for all selected receivers (D-50, SPL, RASTI, STI and Ts, C-80, LF and LFC). The sources used are marked along the top of each of the four graphs. Within each source group the numbers used are marked by a dot pattern see fig.

A

0

5

1

6

2

7

3

8

4

9

TREND3.PLT

An EKgrad plot in showing max value for > 50 ms from the first arrival (high values for < 50 ms are not likely to create echoes, see Prediction theory). If a receiver indicates high values it can be checked in detail by doing a single receiver calculation. PARAM.TXT and PARAMx.TXT also list the max values for > 50 ms.

SPECTRA.PLT

A plot of the SPL octave-band spectra for direct sound, first 50 ms and total as an aid in overall system eq and coverage. Spectra for all receivers are shown in light grey and each receivers data is as plot frame (in a 2DA plot-file) also indicating a line from the sources added to the receiver (dashed if there is no direct sound).

PARAM.TXT

contains source combination/data and results in the form of acoustic parameters. Example:

194

6. SOURCE ADDITION MODULE CATT-Acoustic v8.0a : source add. acoustic param., Copyright © CATT 1988-2007 =================================================================================================== Original file name: C:\CATTDOCS\v8\SHOEBOX_multsrc\OUT\PARAM.TXT Original save time: 2007-03-25 12:11:11 --------------------------------------------------------------------------------------------------Project : v8 Shoebox --------------------------------------------------------------------------------------------------Rec id Location Head direction --------------------------------------------------------------------------------------------------01 -3.00 20.00 1.30 -3.00 19.00 1.30 02 3.00 12.00 1.30 3.00 11.00 1.30 SOURCE COMBINATION --------------------------------------------------------------------------------------------------S# Directivity Aim_x Aim_y Aim_z 125 250 500 1k 2k 4k 8k 16k Delay ................................................................................................... A0 CATT.SD0 1.0 4.7 1.7 70.0 73.0 76.0 79.0 82.0 95.0 95.0 95.0 0.0 B1 CATT.SD1 -2.0 7.7 1.7 70.0 73.0 76.0 79.0 82.0 95.0 95.0 95.0 0.0 S# Hor Ver Roll ...................... A0 -0.0 0.0 0.0 B1 -0.0 0.0 0.0 Direct sound power addition. DIRECT SOUND DELAY/LEVEL (re first arrival) --------------------------------------------------------------------------------------------------S# delay 125 250 500 1k 2k 4k 8k 16k sum 01................................................................................................. A0 9.5 -3.3 -1.1 -1.3 -4.3 -0.1 -0.8 -1.6 -3.1 -1.4 B1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 02................................................................................................. A0 4.5 -1.5 2.0 3.4 2.3 7.6 8.5 9.4 9.9 8.7 B1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Sound from Src B1 arrives first (becomes the reference) to both Rec 01 and 02. Sound from Src A0 to Rec 01 arrives 9.5 ms after Src B1 but 4.5 ms later at Rec 02. PARAMETERS --------------------------------------------------------------------------------------------------Ts [ms] 125 250 500 1k 2k 4k 8k 16k sum ................................................................................................... 01 90.4 79.3 71.9 64.2 56.2 53.9 40.7 28.9 45.4 02 86.5 77.5 72.3 65.9 55.3 54.3 36.8 24.4 39.0 --------------------------------------------------------------------------------------------------D-50 [%] 125 250 500 1k 2k 4k 8k 16k sum ................................................................................................... 01 42.4 49.3 54.5 59.5 69.9 70.8 75.7 90.8 75.3 02 48.6 53.9 57.6 60.7 64.8 66.0 80.0 96.4 80.6

… C-80, LFC, LF, SPL, RASTI, STI Note: results for receivers that are not reached by any direct sound are always hard to interpret since most measures assume that the direct sound is present. In the PARAM.TXT and E_rr.TXT result-files, an absent direct sound will not give a DIRECT SOUND DELAY of 0.0 ms for the sourcereceiver combination concerned even if the source is the closest one to the receiver (had the direct sound not been blocked).

PARAMX.TXT

same as PARAM.TXT but made for MS Excel import (text is within "" and data is semi-colon delimited).

195


196

7. POST-PROCESSING MODULE

7. POST-PROCESSING MODULE The post-processing (PP) module synthesizes binaural as well as other types of room impulse responses (IRs) from generic octave-band echograms created by the Prediction module or by the Source addition module. The module can create multiple source auralization, convolve synthesized or measured impulse responses with anechoic WAV-files, perform relative calibration, as well as file-format conversions and many other supporting functions. The PP module uses settings-files with the extension .PST that store all necessary information to perform a post-processing (except for the echogram-file itself). The settings consist of sample-rate, IR length, output data format, processing methods and more. Several settings-files can be created for different purposes. PST-files can also be used in the Sequence processing module for batch processing. By selecting (Save data for) post-processing in the Full detailed calculation (or for special cases in the Early part detailed ISM dialog) in the Prediction module, files with the name syntax E_ss_rr.ECH are created (ECH is short for echogram). ECH-files are read by the PP module and contain all hall-specific information needed to synthesize a binaural room impulse response (BRIR), as well as various mono, stereo, B-format (for Ambisonic [B-format]) and 5channel impulse responses (IRs) and files for CATT-Walker. The prediction project name is included in the file but it is recommended to keep a copy of the corresponding E_ss_rr.TXT or PARAM_ss.TXT for reference. Note: the more rays that are used the longer the early part will be and more of the unique spatial properties at a receiver position will be included. See Fig 7.1 for an overview of the data-flow for binaural post-processing (source A2 and receivers 01 and 02).

Fig. 7.1 Data flow from prediction to post-processing and auralization, binaural example

The synthesis of a BRIR is explained in the Theory section 7.3. The convolution of a BRIR and anechoic music/speech can be done directly on the PC (a few seconds to a few minutes processing) and the processed sound be listened to using a 16-bit stereo soundcard capable of playing WAV-files - desktop auralization (WAV-files or "RIFF Wave" is part of the Windows Multimedia specification).

197


Optionally Lake DSP CP4 or Huron convolution processors can be used which then enables real-time convolution, head-tracking and virtual reality applications [Lake]. Information on the Lake DSP SIM-file format, which is used to feed the BRIR data to the convolution processors, can be obtained from Huron/CP4 manuals. To limit the number of file-formats used the SIMfile format is used also with software convolution. In the software convolution case all tools necessary are provided and detailed knowledge of the SIM-format is not necessary. The two alternative processing paths are shown in Fig. 7.1. CATT-Walker (documented in its own help-file Walker.PDF) allows for real-time walkthrough based on pre-calculated B-format impulse responses, CATT-Walker is documented in its own help-file Walker.PDF and can be loaded via File|External CATT Tool.

The PP module utilizes HRTF libraries (in the CATTDATA-folder HRTFs), headphone equalization filters (in the CATTDATA-folder HEADPHONES). These are selected in Preferences. Anechoically recorded WAV-files must be placed in the CATTDATA-folder ANECHOIC, a default file to use at convolution can be selected in Preferences

7.1 Menus File|New Settings...

Creates new post-processing settings. If a Template.PST file exists in the CATTDATA-folder it will be used as a template for the created settings-file. File|Open Settings...

Load new post-processing settings (*.PST). File|Save Settings

Saves current settings. File|Save Settings As...

Saves current settings under a new name. Utilities|

Various utilities for file-format conversion etc. Some of the provided functions may also be performed by Lake DSP software, see Lake manuals. The CATT utilities below manipulating Lake SIM-files assume real vectors only (i.e. FIR filters as created by the PP module). The most used utilities have toolbar buttons. Utilities|Information|CATT ECH-file...

Displays essential information from an echogram ECH-file, that is created by the Prediction or the Source addition modules, and that is read by the PP module (reverberation time, prediction project name etc.). Example: Essential ECH-file info -------------------------------------------------------------------------------------Prediction project : Sample project Source location [m] : ( 1.500 3.500 1.500 ) Receiver location [m] : ( 0.000 7.500 1.200 ) Head dir. coord. [m] : ( 1.500 3.500 1.500 ) Detailed refl. to [ms] : 134

198

7. POST-PROCESSING MODULE Early part max order : 44 No of reflections : 19830 (all 8 octaves) + direct Mean free path [m] : 4.19 Mean abs. coeff. [%] : < 15.65 12.95 13.35 15.24 13.20 13.71 : 14.57 14.99 > Reverberation time [s] : < 0.99 1.20 1.15 0.99 1.12 0.96 : 0.65 0.33 > Source gain/eq. [dB] : < 0.0 0.0 0.0 0.0 0.0 0.0 : 0.0 0.0 > Source aux. delay [ms] : 0.0 --------------------------------------------------------------------------------------

Utilities|Information|Lake SIM-file...

Displays the impulse response (IR) file-header of a Lake SIM-file and displays the IR in various ways. The comment section before the actual SIM header is used by the PP module for comments stating the PP version, the prediction project name, initial scale factor needed for calibration, and some other data required for SIM- and WAV-file addition and calibration. General notes: ·

the Play button plays the IR over the soundcard, the complete IR is played independent of Time selected. If the IR is one of a binaural or stereo pair, both channels are played.

Impulse response (IR) :

199


·

the Impulse Response (h) is always scaled to the max sample strength and displayed in % (or not much will be seen), with the dB option the actual levels are shown.

⋅

the dB,Range option displays 10 log10(h )

·

if the Window option is chosen the effect of the windowing as well as the window itself (a raised cosine window) is shown.

2

Frequency Response (FR) :

·

the FR is created by an fft of the time length chosen in the IR and is displayed with all scale factors and eq (Gain_a from the source-file) applied. IRs for normal sources are created re 1m so if only the direct sound is made for a receiver at 1 m from the source on axis (like order = 0 in Early part detailed ISM), if Gain_a is zero or with a natural source, the FR will be at 0 dB as a nice straight line. Consequently, if Gain_a is set at e.g. -10 dB the FR will be at -10 dB and if the Gain_a is set with differing values over the bands it will be a corresponding filter shape.

·

the fft size used is for the closest 2 (the IR is zero-padded as necessary)

·

the Window option smooths the FR. It has most effect when a short time segment is shown.

n

200


Spectrogram (SG) :

·

a Spectrogram is one way to analyze an IR in both time and frequency and to reveal e.g. high-frequency echoes that may not be visible in the bare IR. Most useful time and frequency resolution is ca. 20 ms and 1/3-octaves corresponding well to the human hearing.

·

the SG time length is the same as is chosen for the IR but the Window option does not affect the SG.

·

the time resolution is selected by Time res.: max resolution will give one value per horizontal pixel, max x 3 will give one value for every 3 pixels. Actual time step is displayed above the graph. it may happen, for some combinations of time resolutions and IR lengths, that a time step is too short to evaluate the low frequency contents. The valid frequency range is indicated to the left of the graph with a heavy red line starting at the frequency where the time step corresponds to one period.

·

it may happen with the fixed time resolutions that there should be more than one point per pixel (long IR and short time step) and the results are ambiguous, "(pixel aliasing)" is then shown above the graph.

·

when analyzing the data for each time step, 50% overlap and a Hann window is used.

·

the frequency resolution is selected by Freq res.: max resolution will give one value per vertical pixel, max x 3 will give one value for every 3 pixels. Actual frequency step is

201


displayed above the graph. ·

after a settings-change the SG is first calculated in the background to find current max and min values and the color scale is auto-adjusted (same as using the Auto button). The + and – buttons can be used to manually change the scale e.g. to enhance the later part. Values that are out of range (above the “ceiling” and below the “floor”) get the max and min 2 color values and a white dot. The color-scale is based on 10·log10(|H| ) just like the FR but the level depends on the time and frequency resolution used.

·

if a time step is empty (e.g. when selecting to show a 2 s IR when the actually IR is just 1 s), white is displayed.

Utilities |Information|CATT CWI-file...

Displays the impulse responses contained in a CWI-file for CATT-Walker. It is identical to the dialog for the file-information about a Lake SIM-format file but a selection of receiver is also given. If more details are desired regarding individual positions, they can be post-processed the normal way e.g. with a binaural receiver model. If a CWI is the result of a multiple source CWI-addition, time zero is set to when the sound from the first source arrives. Utilities|Information|Windows WAV-file...

Displays information about a 16-bit WAV-file: If a CATT SCL-file, as created at a software convolution, is present the contents are displayed for reference. If the file is an anechoically recorded file and a CATT AWI-file is found the contents are displayed for reference. Example: 16-bit WAV-file -------------------------------------------------------------------------------------Sample-rate [Hz] : 44100 Type : binaural HRTF-lib : CATT1_PLAIN_44.DAT Headphone : BEYER_DT990PRO_PLAIN_44.DAT Head dir. : 2.500 5.000 -2.500 Receiver : 1 Init Delay : 39.69 ms Aux Delay : 0.00 ms No of samples : 223369 stereo samples Length corresponds to 5.065 seconds @44100Hz -------------------------------------------------------------------------------------D:\temp\V72DOC\ADD\Ftb_a0_01.wav is Anechoic file C:\CATT32\ANECHOIC\FTB_44_AN.WAV (1998-02-19 13:12:12) Convolved with D:\TEMP\V72DOC\ADD\E_A0_01_L.SIM (1999-09-23 08:54:56) and D:\TEMP\V72DOC\ADD\E_A0_01_R.SIM (1999-09-23 08:54:56) Scale-factors for the WAV calibration utility: Filters : 4.09200E+06 WAV-file : 2.83603E-06 Total : 1.16050E+01 (21.3 dB) Not yet calibrated/scaled --------------------------------------------------------------------------------------

The calibration count (last line) gives a measure of the numerical degradation of the file (see WAV-file calibration below).

202


Utilities|Information|HRTF-library...

Displays information about a binary CATT HRTF library. Example:

Description Copyright Subject/Type HRTF-type Normalization FIR-size Sample-rate

: : : : : : :

CATT-Acoustic v7 HRTFs CATT 1997 Unknown single person plain 256 44100Hz

Utilities|Information|Headphone filter...

Displays essential information about a binary CATT headphone filter file. Example:

Description Copyright Subject/Type Left/Right For HRTFs w. FIR-size Sample-rate Description Copyright Subject/Type Left/Right For HRTFs w. FIR-size Sample-rate

: : : : : : : : : : : : : :

Beyer Dynamic CATT 1998 HRS II same filter normalization 4096 44100Hz Beyer Dynamic CATT 1997 Unknown same filter normalization 512 44100Hz

DT990 Pro

plain

DT-990

plain

Utilities|WAV-file Player...

Opens the CATT WAV-file player (can also be made to open automatically after a convolution or a WAV-file addition). The player has three modes.

Single file player

A/B file player

A CATT play-list is a text-file. Example: CATT-PLAYLIST VERSION = 0 C:\Temp\mt_A1_21.WAV C:\Temp\mt_A1_22.WAV "C:\Temp\folder 2\mt_A0_21.WAV" "C:\Temp\folder 2\mt_A0_22.WAV"

203

Play-list player


where "" must be used for filenames with space. CATT-PLAYLIST has to be on line one and the VERSION statement on line two. Note: the free standalone GratisWavPlayer works in a similar way to the built-in one. Utilities|File Scaling...

Enables scaling of either a Lake SIM-file or a Windows WAV-file. The scale factor to be selected must be > 1 and scales the data down. Lake SIM-file (adds factor to text header) (File-type)

Scaling

The selected factor is added to the header so that the calibration utility below can keep track of the total scale-factor. The major reason for scaling a SIM-file is that it may overflow the 24-bit integer convolution in a Lake DSP convolver. An output SIM-file from the PP module is optimally scaled so it should normally not be necessary. For software convolution using the PP module, SIM-file scaling is never required. Windows WAV-file (updates SCL-file if found) (File-type)

Scaling

The software convolver reads and writes WAV-files. The processed sound is auto-scaled for maximum dynamic range. If, however, several receiver-positions in the same hall are going to be compared their relative levels must be correct. With files created by CATT, an SCL-file is also maintained by the scaling and calibration utilities and documents each scaling operation. This utility is not used very often. Utilities|Calibration

Enables relative calibration of either Lake SIM-files or Windows WAV-files (with corresponding SCL-files). Note: Only files created or manipulated by CATT-Acoustic can be correctly calibrated and SIM-file calibration is only necessary for use with a Lake convolver. For details on calibration, see section 7.3.

Lake SIM-files (contained in a single folder) (File-type)

Calibration

This utility performs relative calibration of all SIM-files in the selected folder that matches the specified pattern. Lake SIM-files (across several folders) (File-type)

Calibration

This utility performs relative calibration of all SIM-files in the folders that are listed in the selected list-file and that matches the specified pattern. The list-file is simply a text-file listing the folders with one entry/line. This utility is useful should more source-receiver combinations

204


than allowed for by 260 sources and/or 100 receivers be used. Windows WAV-file (corr. SCL-files required) (File-type)

Calibration

Relative calibration of a set of Windows WAV-format files. The software convolver reads and writes WAV-files. The created WAV-files are auto-scaled for maximum dynamic range. This utility uses the specified WAV-file template and performs relative calibration of all WAV-files in the same folder that have SCL-files stating exactly the same anechoic source-file as in the template. The SCL-file is updated so the same WAV-file can be calibrated again against other files. If the scale-factors of the WAV-files are very different, repeated calibration may cause noticeable numerical degradation. The SCL-file therefore also keeps track on the calibration count as a measure of the degradation (quantization noise). The Utilities|Information| Windows WAV-file utility above reads and displays the contents of an SCL-file if present. Note: to use the WAV-file addition utility, the WAV-files do not have to be calibrated in advance since the scale-factors are applied at addition. WAV-file calibration can also be performed by the Sequence module, Section 8. Utilities|SIM-file addition

Adds created SIM-files to enable multiple source auralization where each source sends out the same source material (only available in the full auralization version). The result SIM-file can then be convolved with anechoically recorded material in the usual manner. To make SIM-file addition meaningful, the SIM-files must be compatible with each other. To determine the compatibility the filename syntax E_ss_rr_f.SIM is relied upon and also that the SIM-file header has not been altered after post-processing (other than by CATT utilities described in this section). The following properties are checked: ·

same receiver-model (for binaural also the headphone eq and HRTFs must be the same)

·

same receiver position (file-name check)

·

same sample-rate

·

same head-direction (note that head-direction source cannot have been used at prediction since the head-directions then will be incompatible)

If in doubt about compatibility, use the Utilities|Information menu to display the SIM-file header. Gain

Impulse response (SIM-file) addition

Additional wideband gain for the selected SIM-file. Aux delay


Auxiliary electrical delay.

205

7. POST-PROCESSING MODULE Update


Updates the selected SIM-files with gain and delay settings. If Enter is pressed when gain or delay edit boxes are active it performs the same action. The list-box displays all selected files (only the base-name where e.g. for binaural _L.SIM and _R.SIM will be appended at addition). It also displays the initial delay (source-receiver delay, ms), current gain (dB), and auxiliary delay (electrical, ms) for each file. Add first SIM-file/Add SIM-file(s)


Displays a standard Open File dialog. The first file selected determines the compatibility. Subsequently, only compatible files and such that are not already listed can be added. It is sufficient to select e.g. the left file (E_ss_rr_L.SIM) for binaural and e.g. the omni (E_ss_rr_W.SIM) file for B-format. When the actual addition is performed all files required for a format will be used. Several files can be selected in one operation by using multiple nd nd rd selection in the Open dialog (Shift-LMB, Ctrl-LMB etc.). Note: If 2 (R,S,T,U,V) or 2 and 3 order (R,S,T,U,V,K,L,M,N,O,P,Q) SIM-files are found they are also added and the sum PLTst nd nd rd file created shows all 1 and 2 order components (when adding 2 and 3 order files, either the W, X, Y or the Z file has to be selected, rather than one of the higher order components). Remove


Remove the currently selected file from the list-box. Result SIM-file base-name


The base-name of the resulting SIM-file. _M.SIM, _L/_R.SIM, _W/_X/_Y/_Z.SIM (and higher order components if available), or _1/_2/_3/_4/_5.SIM_will be added depending on receiver model. OK


Performs the addition and creates the result SIM-file as well as a plot-file with the same basename as the result-file and with _I.PLT added (e.g. E_01_I.PLT). This 2DA plot-file shows plot-frames indicating the direct sound of each source as well as file-names, Gain and Auxdelay of each added file. With B-format addition, if 3rd or 2nd order SIM-files are found added st nd the sum PLT-file created shows all nine 1 and 2 order components (any 3rd order components are added but not plotted). Note: with binaural SIMs, due to headphone eq, a delay may exist so that the actual direct sound comes a few ms after the dotted line. This delay has no audible effect since it is contained in all responses added.

206


Utilities|WAV-file addition

Adds created stereo/binaural WAV-files to enable multiple source auralization but with possibly different anechoic source material for each source (only available in the full auralization version). To make WAV-file addition meaningful, the WAV-files must be compatible with each other. To determine the compatibility an accompanying SCL-file is required (created at convolution). The following properties are checked: ·

same receiver-id

·

same receiver-model (stereo or binaural, for binaural also the headphone eq and HRTFs must be the same)

·

same receiver position (info saved in SCL-file created at convolution based on SIM-file name)

·

same sample-rate

·

same head-direction (note that head-direction Source cannot be used at prediction since the headdirections then will be incompatible.

If in doubt about compatibility, use Utilities|Information that can display WAV/SCL-file headers. WAV-files do not have to be relative calibrated to take part in addition since all scalefactors are applied properly at addition. The addition is documented in a text-file (.TXT) with the same name as the created sum WAV-file (which WAV-files have been added etc.) so that operations can be tracked. Note: for addition of B-format or 5-channel use the MultiVolver that will convolve and add in one step. Gain

Binaural/Stereo WAV-file addition

Additional wideband gain for the selected WAV-file. A major problem with WAV-file addition is that the scale-factors in most anechoic WAV-files are unknown. The Gain factor can be used to compensate so that if it is known that a weak signal has been amplified e.g. 10dB before sampling as compared to a strong signal, the weak-signal WAV-file must be assigned a Gain of -10dB so that the two anechoic signals will be compatible in level. The addition is made using floating point so there is no major numeric loss in adding a weaker signal to a stronger. Note: v8.0 does not use scale information in anechoic WAV-info files (AWI-files). Aux delay


Auxiliary electrical delay. Update


Updates the selected WAV-file with Gain and Aux delay settings. If Enter is pressed when Gain or Aux delay edit boxes are active it performs the same action. The SIM and WAV scale factors are both automatically taken into account when adding.

207


The list-box displays all selected files. It also displays the initial delay (source-receiver delay), current gain, and auxiliary delay for each file. Add first WAV-file/Add WAV-file(s)


Displays a standard Open File dialog. The first file selected determines the compatibility. Subsequently, only compatible files can be added. Several files can be selected in one operation by using multiple selection in the Open dialog (Shift-LMB, Ctrl-LMB etc.). Remove WAV-file


Remove the currently selected file from the list-box. Result WAV-file name


The full name of the resulting WAV-file. Auto play after addition


If selected, the resulting sum WAV-file is played automatically after completed addition using the CATT WAV-file player. OK


Performs the addition and creates the result WAV-file. Some notes on the addition with a schematic explanation with 6 files to be added:

i a 1, 2, 3,... s 1 2 3 4 5 6 ->

initdelay auxdelay data in file 1, 2, 3,… sum of files 1-6

iiiiiiiiiiiaaaaaaaaaalllllllllllllllllllllllllllllllllllll iiiaaaaaaaaaa2222222222222222222222222222222222222 iiiiiiiiiiiiiiiiiiiiiaaaaa3333333333333333333333333333333333333 iiiiiiiiiiiaaaaaaaaaa444444444444444 iiiiiiaaaaaaaaaa555555555555555555555555555 iiiiiiiiiiiiiiiiiiiiiaaaaa66666666666666666666666 sssssssssssssssssssssss

i.e. the new sum WAV-file will start at t_start= minimum(initdelay+auxdelay) time (i.e. determined by file 2) and last to t_end= minimum(initdelay+auxdelay+datalength) time (i.e. determined by file 4) and be t_end-t_start long. It is thus of interest to have all WAV-files of similar lengths (delays taken into account) to get a long enough common sound to listen to. Sample use of this utility is binaural auralization of stereo replay: 1) Place two sources on stage, one for the left channel and one for the right channel; 2) Predict and post-process both sources using a head-direction typically in-between the two speakers; 3) Convolve the left speaker impulse responses with the left channel of the anechoic WAV-file and then the right response with the right channel of the anechoic file; 4) Select WAV-file addition and add the two WAV-files and binaural auralization of stereo replay is created.

208


The initial delay (source to receiver) and any extra (aux) delay will be taken into account and just like for SIM-file addition (that assumes the same anechoic material for all sources) the aux delay can be altered. It is most natural to use the WAV-file addition utility with several sources using different material (e.g. stereo, 5-channel (see below) replay, or speech+noise) but it can also be used to add WAV-files using the same anechoic material just like the SIM-file addition (but it is less efficient). Note: stereo to binaural can also be performed by the MultiVolver using a 2x2 setup. Utilities|CWI-file Addition...

Analogous to SIM-file addition but for CATT-Walker. A text-file is created with the same name documenting the addition. No matter how many CWIfiles (sources) are added, roughly the same amount of CATT-Walker CPU is required as for a single source. Note that for CWI-files to be added, they all have to be created from ECH-files saved at the same time, this is to ensure that added CWI-files are compatible. Also note that to resolve source interference at e.g. 1 kHz you need to have a receiver spacing of half a wavelength or less at 1 kHz (17 cm) so don’t over-interpret interference effects (or lack thereof) for big receiver spacings. A similar caution is advised regarding fine directivity details, to be resolved a sufficient receiver density is required.

209


Utilities|File conversion

16-bit INT -> Mono WAV

File conversions

Conversion from a mono "signed 16-bit integer" to Windows WAV-format (extension is changed to .WAV). This utility may come in handy when converting anechoic material from other sources (many sound-file conversion programs can create header-free signed 16-bit integer files). Also the sample-rate in Hz needs to be specified. MATLAB MAT -> Lake SIM

File conversions

Conversion from MATLAB v4 .MAT [MATLAB] to Lake SIM-format (extension is changed to .SIM). The data has to be properly scaled since the Lake-format files are 32-bit integers. For Lake convolution only 24 bits or less can be used (see Lake manuals), for software convolution the full 32 bits can be used. Note: this utility only reads the first variable in the file - save only the IR from MATLAB or save it as the first variable. Lake SIM -> MATLAB MAT

File conversions

Conversion from Lake SIM-format to binary MATLAB v4 MAT-format (extension is changed to .MAT). The MATLAB variable name of the vector will be sim_data. If the sample-rate is contained in the SIM-file header, it is named sim_fs. Lake SIM -> MLSSA TIM

File conversions

Conversion from Lake SIM to a binary MLSSA TIM-file [MLSSA] (extension is changed to .TIM). Also the sample-rate in Hz needs to be specified. The sample-rate may be defined in the SIM-file header. If it is, it is compared to the given value. Lake SIM -> Windows WAV mono

File conversions

Conversion from Lake SIM to Windows WAV mono (extension is changed to .WAV). The sample-rate may be defined in the SIM-file header. If it is, it is compared to the given value. This utility enables direct listening to a IR (equivalent to first convolving it with a unit impulse and then listen), see note below. Lake SIM -> Windows WAV stereo

File conversions

Conversion from Lake SIM to Windows WAV stereo (extension is changed to .WAV). The sample-rate may be defined in the SIM-file header. If it is, it is compared to the given value. Left and right SIM-files, output WAV-file name, and the sample-rate needs to be specified. This utility enables direct listening to a BRIR (equivalent to first convolving it with a unit impulse and then listen). A SIM-file can also be listened to directly via Utilities|Information| Lake SIM-file.

210

7. POST-PROCESSING MODULE MLSSA TIM -> MATLAB MAT

File conversions

Conversion from MLSSA TIM to MATLAB v4 MAT-format (extension is changed to .MAT). The MATLAB variable name of the data will be tim_data and the sample-rate in Hz will be tim_fs. The data can optionally be saved as double precision floats instead of single precision. Note: only a PC MATLAB can read the single precision format. Windows WAV mono -> MATLAB MAT

File conversions

Conversion from Windows WAV mono to binary MATLAB v4 MAT-format (extension is changed to .MAT). The MATLAB variable name of the vector will be wav_data and the sample-rate in Hz will be wav_fs. Windows WAV mono -> Lake SIM

File conversions

Conversion from Windows WAV mono to Lake SIM-format (extension is changed to .SIM). Windows WAV stereo -> Lake SIM

File conversions

Conversion from Windows WAV stereo to two Lake SIM-format (one SIM-file/channel, _L.SIM and _R.SIM will be added to the name and extension is changed to .SIM). MLSSA TIM -> Windows WAV stereo

File conversions

Conversion from two MLSSA TIM to Windows stereo WAV-format (extension is changed to .WAV). This utility enables direct listening to a measured BRIR using a sound-card. Left, and right TIM-file and the output WAV-file names need to be specified. Note: the sample-rate has to be one supported by Windows soundcard (typically 11025, 22050 or 44100 Hz) or it may not play correctly. Utilities|SIM-file Filtering...

A general FIR-filtering tool for SIMfiles (only available in the full auralization version). Typical use is to apply a loudspeaker on-axis filter, band-limit the response, or apply a head-phone eq (for binaural SIM-files where no eq was made during processing i.e. the headphone "(none)" was selected in Preferences). This tool is quite open and general and not a lot of checks are performed on the validity of a filtering action (mainly that samplerates are compatible for TIM and Headphone eq). Input SIM-file

SIM-file filtering

SIM-file to be filtered. FIR filter

SIM-file filtering

FIR-filter file-name. The file can be in the following formats: MATLAB v4 (.MAT), MLSSA (.TIM) or a CATT headphone eq filter (.DAT, assumes that left and right filters are identical).

211

7. POST-PROCESSING MODULE Output SIM-file

SIM-file filtering

SIM-file to be created as a result of filtering. Truncate output to input length

SIM-file filtering

If a vector length N is filtered with an FIR-filter length M the resulting length is N+M-1. This option makes also the output be N long by skipping the M-1 last samples. For a typical long decaying SIM-file and a short filter this truncation will not be audible. Normalize filter to unit power gain

SIM-file filtering

Typically, filters used should have the proper normalization but this option gives a reasonable overall gain change by normalizing the filter to unit power gain. Utilities|SIM-file Sample-rate Conversion…

A tool that can samplerate convert SIM-files (only available in the full auralization version).

Input SIM-file

SIM-file sample-rate conversion

SIM-file to be converted. Sample-rate in


Sample-rate of the file to be converted. Output SIM-file


SIM-file to be created as a result of conversion. Sample-rate out


Sample-rate of the converted file. The OK button is disabled for not supported conversions. Conversions implemented are: From/To

16000 22050 32000 44100 48000 88200 96000

16000 -

22050

32000 x

x

44100

48000 x

88200

96000

x -

x

x

-

x

x -

x x

x

212

x -

7. POST-PROCESSING MODULE Utilities|Calculate IACC...

Calculates the Inter-Aural Cross Correlation Coefficient for a binaural SIM-file pair. Select one of the files in the pair (left or right) and the IACC will be displayed, See Theory for details. As a General settings option, IACC can also be calculated after completed post-processing and be put in the SIM-file header. Note: the responses are assumed to have no significant common initial delay (i.e. no initial zeros) since that is how CATT SIM-files are created. Utilities|Crosstalk Cancelling...

Crosstalk cancelling enables loudspeaker listening to binaurally recorded or simulated sound. The background is that when listening to a binaural recording over loudspeakers (in a typical symmetrical stereo configuration), the left as well as the right ear is reached by sound from both left and right speakers while for binaural replay the left signal is supposed to be sent only to the left ear and the right signal only to the right ear. The crosstalk cancellation technique uses the HRTFs for the angle to the loudspeakers from the listening position and creates filters that cancels the "error" signal i.e. the signal from left to right and right to left. If this can be done perfectly, one will hear the same sound as one would in headphones but there are several problems that need to be addressed. The implementation here uses the so called "Shuffler" filter configuration and a more thorough description and discussion is found in the Theory section. The description below concentrates on the interface.

Spacing, S

Crosstalk cancelling for loudspeaker replay

Loudspeaker spacing as measured between the acoustic centers. Distance, D


Loudspeaker distance as measured between the center of the head and the line joining the two loudspeakers.

213

7. POST-PROCESSING MODULE Apply loudspeaker correction


Optionally uses two supplied inverse filters to correct the on-axis response of the loudspeakers used. Files must be Finite Impulse Responses (FIRs) in MATLAB v4 PC format and be created for the samplerate used, see Theory. There is no practical limit on the length of these filters. (Crosstalk option)


The Crosstalk cancellation can operate in three basic ways: Normal operation operates on a binaural room impulse response SIM-file pair (without headphone eq). The new SIM-file pair will have its first filename character changed to "X" (E_ss_rr_L.SIM becomes X_ss_rr_L.SIM) and the "HEADPHONE EQ=" fields in the SIM-file headers are changed from "(none)" to "TRANSAURAL" so that pure binaural and transaural cannot accidentally be mixed in SIM- and WAV-file additions). The new SIM-file pair can then be convolved with anechoic sound in the normal manner but the resulting WAV-file is to be played over loudspeakers instead of over headphones. If the same basic response is to be used for headphone replay the original SIM-file pair can be filtered for headphones using the SIM-file filtering utility or the Sequence processing module. In this operational mode, the samplerate is given by the SIM-file pair. Left (or Right) channel calibration creates calibration filters for finding and evaluating the sweet-spot, see Theory. When convolved with e.g. pink noise they will give a left (or right) calibration signal. If the filters work well and the listener is in the sweetspot, sound should appear to come only in the left (or the right) ear. A good way to find the sweet-spot is to cover the left ear and move the head sideways until a minimum in the right ear is heard. When the left ear is uncovered the sound should be as if it was played in the left headphone only. In this operational mode the desired samplerate must be specified and the files created are calib_L.SIM and calib_R.SIM where calib is the given base-name. Create X-talk filters only creates dedicated crosstalk cancellation filters that can be used to process e.g. binaural recordings. In this operational mode the desired samplerate must be specified and the files created are xtalk_L_L.SIM, xtalk_L_R.SIM, xtalk_R_L.SIM and xtalk_R_R.SIM where xtalk is the given base-name. To crosstalk cancel a binaural recording the simplest is to use the MultiVolver with a 2x2 setup.

The dummy head should typically have had microphones at the blocked ear canal entrance but also simple clip-on binaural microphones can give excellent results. (Filter smoothing)


Various options for smoothing affect how well the cancellation works and the size of the “sweet-spot”. See Theory for a description of the filters S and A and further details. HRTFs

smoothes the S and A (HRTF) magnitudes before creating the crosstalk cancellation filters. Makes the left and right HRTFs have minimum phase with an added pure delay to A. In general, smoothing the HRTFs affects the crosstalk cancelling ability too much.

Filters

smoothes the magnitude of the S+A and S-A filters and forces them to minimum phase before inverting. In general, smoothing the filters improves the crosstalk

214


cancelling ability at the expense of affecting the direct signal for the higher frequencies (i.e. left to left and right to right). Recommended settings are 1) both options set to None or 2) HRTFs set to None and Filters set to 1/3-octave smoothing (or finer). Samplerate


The desired samplerate for calibration and dedicated crosstalk filters. For Normal operation the samplerate is taken from the SIM-file pair given. 10kHz LP-filter


Applies a 10 kHz low-pass filter to the resulting crosstalk cancelled filters. This has the effect of making the localization less sensitive to head rotations and widens the sweet-spot somewhat at the expense of some high frequency loss. SIM-file name


Name of the SIM-files to read and/or create. For Normal operation either the left SIM-file (e.g. E_A0_01_L.SIM) or the right (E_A0_01_R.SIM) is selected and (for the example) X_A0_01_L.SIM and X_A0_01_R.SIM are created, i.e. the first letter is changed to X. Note: to enable cross-talk cancellation of files created for HeadScape, an exception is if the first letter is H where the old files instead are overwritten (make a copy in another folder before processing). The reason is that SIM-files created for HeadScape have the syntax HhhhvvvL.SIM and HhhhhvvvR.SIM and HeadScape cannot load files if they start with X. If files are created for HeadScape (but without headphone eq) followed by a batch crosstalk cancellation in the Sequence module, the tracking mechanism can be used to rotate the sound around the head with a transaural replay. If nothing else, this is a great way of testing the head-tracking and cross-talk operation. Save


Saves the current settings. If crosstalk cancellation is applied using the Sequence processing module the latest saved settings are used. OK


Performs the requested processing and saves the settings. Cancel


Leaves the dialog without saving the settings (unless Save has been pressed). Utilities|Make Anechoic WAV-file Info…

Compilation of an anechoic WAV info file (AWI-file) for an anechoic recording. The information is used e.g. when the Post-processing:Utilities menu for file information is selected and it can also be used while specifying data in the source-file. For future revisions, this calibration info may also be used with WAV-file addition.

215


Anechoic file

Compile an anechoic WAV info-file

The anechoic WAV-file to compile the info-file for. Distance (Anechoic recording)


The distance that the recording was made at, measured from the acoustical center of the source. Pre-amp (Anechoic recording)


Assuming that, for example, the calibration file (see below) is recorded with a setting of 0 dB on the microphone pre-amp and the recording is made with a setting of 10 dB, 10 dB should be entered here. The reason for this option is that a fixed-level calibration signal may have been made that is stronger or weaker than the actual material recorded thus forcing a preamplification change to record to full scale on e.g. a DAT or a PC sound-card. Source (Anechoic recording)


For example "Guitar", for reference purposes. Description (Anechoic recording)


A more detailed description, for reference purposes. For extended information, place the description in a file and enter the filename. Copyright (Anechoic recording)


For reference purposes. Measured SPL (Calibration)


The measured SPL of the calibration signal (e.g. a loudspeaker sending out pink noise). This

216


info is used to calculate the Lp1m in octave-bands of the source. Note: if no calibration signal has been measured but the level of the source at 1m on axis is known anyway (or can be estimated) the anechoic WAV-file itself can be given as the calibration file together with the known value at e.g. 1 kHz. Calibration file (Calibration)


A WAV-file (at the same sample-rate as the recording) containing the calibration signal at a known level. Name (Recommended directivity)


If a directivity is known for the source in question, it should be entered here. OK


Creates the WAV info-file. The filename will be wavname.AWI in the CATTDATA-folder ANECHOIC assuming that the name of the WAV-file is wavname.WAV). If calibration info is given, the SPL at 1m is calculated as an average using a Hanning window with 50% overlap sliding over the data in the calibration file and the anechoic file. Also standard deviations are displayed to give an idea of how much the level varies during the sample. The following Utilities menu items are only available in special full versions with add-ons for Lake DSP convolution processors [Lake] or for users that are about to measure HRTFs or headphone eq., available on request. Utilities|Create Sphere HRTFs...

This item creates analytical HRTFs from a solid sphere that can be compiled into a CATT-format HRTF-library (see below). A short text-file describing the HRTFs is created in the selected HRTF folder (name HRTF_info.TXT). HRTF folder

Create individual sphere HRTFs

Folder for the HRTFs to be created. Sphere diameter (HRTF information)


Diameter of the solid sphere.

Sphere ear angle (HRTF information)


Angle from the front to the point on the sphere surface to act as ear, see figure (top view):

FIR size (HRTF information)

angle front


Size of the created FIRs. For better results use long filters.

217

7. POST-PROCESSING MODULE FIR sample-rate (HRTF information)


Sample-rate for the created HRTFs. Frontal normalization (HRTF information)


Makes HRTFs be frontal flat by dividing all HRTFs by the frontal ones. To be used with headphone eq normalization “frontal”. Without this option checked the headphone eq type must be “plain”. When compiling an HRTF library from the created filters the normalization type most be set according to this choice (see the created HRTF_info.TXT). Utilities|Compile HRTF-library...

This item compiles individual measured or analytical HRTFs into a CATT-format HRTF-library. With the more general handling of HRTFs introduced with v8.0c, libraries can not be created from within CATT-Acoustic, this utility will only create the previous fixed 15° resolution. To convert measured HRTF data in other resolutions and for further details on HRTF library creation contact CATT. Libraries are created in the CATTDATA-folder HRTFs folder and can then be selected in Preferences.

HRTF MAT-files folder

Compile HRTF-library

Folder where the HRTFs in MATLAB v4 format are to be read. Type (HRTF information)


sphere

HRTFs originate from an analytical sphere approximation

artificial head

HRTFs originate from a measurement on an artificial head

person

HRTFs originate from an individual

218

7. POST-PROCESSING MODULE person average HRTFs originate from some kind of average over several persons simple ITD/ILD

HRTFs are simple Inter-aural Time and Inter-aural Level Differences

other

HRTFs are of any other kind (specify in Description)

Sphere diameter (HRTF information)


If Type = sphere : Diameter of the solid sphere. To be set the same as when the sphere HRTFs were created (see the created HRTF_info.TXT). Sphere ear angle (HRTF information)


If Type = sphere : Angle from the front to the point on the sphere surface to act as ear. To be set the same as when the sphere HRTFs were created (see the created HRTF_info.TXT). Normalization (HRTF information)


What kind of normalization the HRTFs have: plain

the HRTFs are as directly measured at the blocked ear-canal entrance

frontal the HRTFs are measured at the blocked ear-canal entrance but all filters are divided by the frontal filters (requires frontal filters to be included in the headphone eq) diffuse the HRTFs are measured at the blocked ear-canal entrance but all filters are divided by an average "diffuse-field" HRTF. Requires the headphone used to have a corresponding diffuse-field eq. other

Any other measurement type (e.g. measurements at the ear drum). Use Description to document. This will require special headphone eq filters measured to fit to the particular normalization or HRTF measurement made.

Sample-rate (HRTF information)


Sample-rate of the FIRs specified in the MAT-files and of the created library. FIR-size (HRTF information)


Size of the FIR-filters to be used for the library. Must be the same size or larger than the FIRs specified in the MAT-files. Library name


Library name (max 28 characters since a string specifying normalization and sample rate is added). Example: for 44100 Hz and plain normalization the full name will be name_plain_44.DAT. Description (Library information)


Library description (max 30 characters) to be used e.g. for the name of the person or artificial head whose HRTFs have been measured. For reference purposes only. Copyright (Library information)


Name of the HRTF copyright holder and other similar information.

219

7. POST-PROCESSING MODULE Subject/Type (Library information)


Name of the measurement subject such as an individual’s name or type of artificial head. Utilities|Make Headphone Filter...

This item creates a CATT-format headphone eq filter from MATLAB v4 MAT-files. The resulting filter is placed in the CATTDATA-folder HEADPHONES and can be selected in Preferences for use at processing. Filters are assumed to have been measured at the blocked ear-canal entrance unless normalization type other is specified. Details on creating new filters are supplied only with the special Huron add-on version, available on request.

FIR MAT-file

Make headphone filter

Name of filter-file in MATLAB v4 format. Separate left and right filters


Uses separate left and right headphone filters. Note that this option will make the eq be tightly coupled to a particular set of HRTFs or a particular pair of headphones that has be found to have different left/right responses. For HRTFs with normalization (Filter information)


Specify what type of HRTF normalization the filter is made for. Unless the type other is used HRTF and filter normalization cannot accidentally be mixed. Sample-rate (Filter information)


Sample-rate of the MAT-file filter and the created CATT-format eq file. FIR size (Filter-information)

Make Headphone filter

Size of the equalization FIR-filters. Must be the same size or larger than the FIR(s) specified in the MAT-file(s).

220

7. POST-PROCESSING MODULE Headphone (Filter-information)


Brand name and type of headphone. Copyright (Filter-information)


Name of company that holds the copyright of the measurement and similar information. Subject/Type (Filter-information)


The subject used for the eq measurement e.g. an individual's name or type of artificial head. Filter name (Filter-information)


Descriptive headphone name e.g. "BRAND_TYPE". Max 28 characters, normalization type and sample-rate will be added. Utilities|5-Ch Binaural Setup...

This is a very special option and should only be used if the steps involved are fully understood. It is not meant for general-purpose room acoustic simulation but for research and experimentation with various recording and replay techniques. Basically, it creates binaural IR SIM-files as from five surround speakers.

Distance

Anechoic 5-channel binaural setup

Distance from the listening position (receiver 01 placed at 0;0;0). Angle


Angle from the negative y-axis. Symmetric setup


Makes R1 and R2 be mirror copies of L1 and L2 and forces C0 to be at x = 0.

221

7. POST-PROCESSING MODULE Include initial delays in response (for Lake convolution) Anechoic 5-channel binaural setup

The distances to the speakers affect the initial delay and 1/r attenuation but all responses normally start at time 0 and the initial delay is included in the SIM-file header just as for normally created SIM-files. With WAV-file convolution/addition and SIM-file addition the delays are taken into account. With Lake convolution the delays will have to be added by DSP unless this option is checked. The different initial delays are then taken into account by shifting the responses in time so that only the response of the closest speaker starts at time zero. Create GEO-files for non-anechoic setup


Optionally prediction project-files are created: MASTER.GEO, SRC.LOC, REC.LOC, Speakers.PRD that can be used as templates to create a non-anechoic 5-channel simulation either by using the simple shoebox shape with diffusing walls (see instructions in the created GEO-file) or by using the receiver layout for an existing room model. Speakers.PRD: Sources C0,L1,L2,R1,R2 represent speaker positions describe below, receiver 01 represents the listening position. Sample-rate


Sample-rate of the created SIM-files. Result folder


Folder were all created files are placed. OK


5 speakers, all pointed at an assumed listener and 5 SIM-file Left/Right pairs, containing the HRTFs concerned, are created in the selected folder:

E_C0_01_L.SIM E_C0_01_R.SIM

Front Center

E_L1_01_L.SIM E_L1_01_R.SIM

Front Left

E_L2_01_L.SIM E_L2_01_R.SIM

Back Left

E_R2_01_L.SIM E_R2_01_R.SIM

Back Right

E_R1_01_L.SIM E_R1_01_R.SIM

Front Right

C0 L1 x

R1 01

L2

R2

y

The name syntax is the same as for normally created binaural impulse response SIM-files and can be used as such (SIM-file addition and the WAV-file addition after convolution). In addition a plot-file Ch5.PLT is created showing the speaker layout. SIM-files created are optimized in level for Lake convolution (makes no difference for software convolution). The goal is to binaurally simulate 5 ideal loudspeakers as placed in a perfect anechoic chamber. The main use of this option is together with a Lake Huron to be able to listen to a 5channel recording (e.g. Dolby Surround) binaurally.

222


With software convolution, a mono WAV-file for each of the 5 channels can be convolved with the corresponding filters and the 5 resulting stereo WAV-files be added together (the WAV-file addition utility). After convolution the WAV-files can be calibrated only if all 5 channels use the same anechoic material. WAV-file addition can be used to create a binaural replay of a simulated 5-channel system. Non-symmetrical setups are allowed so that, for example, the effect of non-symmetrical speaker placement can be studied. For the HRTFs and the headphone eq the Preferences settings are used just like for normal post- processing and are documented in the SIM-file headers. Actual processing is simplest to perform in the MultiVolver using a 5x2 setup a.k.a. binaural down-mix. If a non-anechoic replay is desired (e.g. a living-room using another configuration of MASTER.GEO) a full prediction of C0->01, L1->01, L1->01, L1->01, L1->01 followed be binaural post-processing must be made so that new E_C0_01_L.SIM etc. will be created for a non-anechoic 5-speaker setup and a 5x2 process can be run again. To compare the 5-channel recording to a full binaural simulation (i.e. an artificial head recording, use A0->01 (head-direction source) but post-process binaurally and convolve with the anechoic material.

223


7.2 Dialogs

The main Post-processing dialog controls settings for the combination of actions to be performed, which types of data-files to create, and software convolution options. For research and other special uses, it is possible to create only the early, or only the late part. For normal auralization both the early and the late part should of course be included. When Save and Run... is clicked the current PST settings-file is saved, and an ECH-file to post-process must be selected. The PP module then performs the actions requested.

224


General settings The General settings dialog controls basic settings such as receiver model, length and sample-rate of the room Impulse responses (IR) filters to be created.

Length (Impulse response (IR))

General settings

Length in ms (per channel) of the total synthesized IR in Lake SIM-format. Each time this item is altered the corresponding IR size in points is updated. Recommended length corresponding to a 60 dB decay (at least a 40 dB decay i.e. 2/3 of the reverberation time or the truncation is clearly audible). If the low frequency T-30 is much longer than at mid frequencies, choose a Length according 60 dB for the mid frequencies. Samplerate (Impulse response (IR))

General settings

Sample-rate (sampling frequency) of the synthesized and saved IR in Lake SIM-format. Each time this item is altered the corresponding IR Size is updated. For software convolution 44100 Hz is recommended, while for Lake convolution 48000 Hz is used with analog I/O and with digital I/O it depends on Huron settings (typically 44100 Hz). Transform-size (Reflections)

General settings

Transform-size for use in digital octave-band filters, spline, Hilbert transform, IFFT, FFT etc. for synthesizing reflections in the IR. Each reflection path transfer function will be approximated with half this number of points ranging from 0 to the Nyqvist frequency. Recommended sizes: Rate, Hz

Size

44100

512 (lower not allowed)

48000

512 (lower not allowed)

225

7. POST-PROCESSING MODULE Apply source eq/gain (Reflections)

General settings

The normal case is to include the source eq/gain (based on Gain_a from the source-file) so that the created impulse responses are affected by the system eq/gain for an electroacoustical source (it has no effect for a natural source). If this option is not checked, it is as if the acoustical input was through a system with unit gain but with the directivity of the loudspeaker selected. See Section 2.3 and 7.3 for details. Fractional delay (Reflections)

General settings

Fractional delay means that a reflection is not added at the nearest sample bin but an interpolation technique is used to add it at the exact "fractional" delay. Slightly longer processing but useful for some special applications. (Stopband suppression)(Reflections)

General settings

How much to suppress above the upper limit of the highest octave-band and optionally below the lower limit of the 125 Hz band (88 Hz). These settings affect only direct sound and firstorder specular reflections that are constructed using a spline magnitude, see Fig 7.3 in Section 7.3. Level (Stopband suppression) (Reflections)

General settings

The suppression is made using a raised cosine bell from the lower limit of the 125 Hz octaveband to 0 Hz. Recommended value -40 dB. Suppr. below 88 Hz (Stopband suppression) (Reflections)

General settings

How to handle the low frequency part of the reflection path transfer function amplitude (below the lower limit of the 125 Hz octave-band: 88 Hz). Suppresses the low frequencies using a raised cosine bell function 88 to 0 Hz. If "off" a flat response down to 0 Hz is created for each synthesized RP impulse response. LF suppression should in most cases be off since it requires large transform-sizes to work well. Recommended setting off. (Receiver-model)

General settings

Type of directional response to use for the receiver. Omni-directional

Apply no receiver directivity at all. Creates a mono omni-directional response (same as using a (1.0;0) Mono microphone). _M is added to the ECH-file base-name.

Binaural

Apply measured HRTFs. Headphone equalization and HRTF library are selected in Preferences (Prefs button). IACC can optionally be calculated and will be put in the SIM-file header (use Utilities| Information|Lake SIM-file to check the values) _L and _R are added to the ECH-file base-name.

Mono microphone

Mono microphone directivity. The microphone direction is the same as the head direction selected at prediction. See Mic settings below. _M is added to the ECH-file base-name.

Blumlein/X-Y stereo

Create a two-channel conventional coincident stereo IR where Mic settings and Angle between the mics can be selected. _L and _R are added to the ECH-file base-name.

226

7. POST-PROCESSING MODULE Mic settings (Receiver-model)

Pre-made types available are: Omni, Cardioid, Hyper-cardioid, Super-Cardioid, Figure-of-8. For User defined and further details see Section 7.3. B-format (WXYZ ) (Receiver-model)

B-format [B-format] is mainly intended for use with CATTWalker, a Lake Huron workstation running AniScape or a Convolver connected to a Speaker Decoder or a Binaural rd Decoder but also the The FIReverb Suite (MultiVolver) or 3 party convolvers/decoders can be used. _W, _X, _Y and _Z etc. are added to the ECH-file base-name Order (Receiver model) st

1.

Creates four different responses according to 1 order B-format as if measured by four microphones where X, Y, and Z are figure-of-8 types pointing in X, Y, and Z directions (X is at the front i.e. same as the head-direction) and W is an omni-directional response st divided by 1.414. For CATT-Walker choose 1 order.

2.

Creates 1 and 2 order components WXYZ RSTUV.

3.

Creates 1 , 2 and 3 order components WXYZ RSTUV KLMNOPQ.

st

st

nd

nd

nd

rd

rd

For 2 and 3 order, see Theory. nd

Note: CATT tools only utilizes WXYZ so external software is required to utilize 2 or higher order. The exception is SIM-file addition that will add also higher order components if found and PureVerb ≥ v3.2e can convert WXYZ + RSTUV + KLMNOPQ to WAV-format with retained relative channel calibration. AniScape (Receiver model) st

(Full Huron version) When checking this button enabled with 1 order B-format, the following is performed creating all required files for Lake AniScape: • output samplerate is the selected one divided by 3. For Huron digital I/O typically choose 44100 Hz (giving reverb processing at 14700 Hz) and for Huron analog I/O choose 48000 Hz (giving reverb processing at 16000 Hz). • reflection orders lower then 2 are excluded (Early part settings) • a GEO-file for Lake AniScape is automatically created along with the SIM-files. The filesname syntax is E_ss_rr_.GEO and the file is created in the same folder as the SIM-files.

The created GEO-file can be selected in the Huron Sound Field Filter VRack application for real time direct sound and first order reflections + a reverberation valid for the room model in question (that uses the names E_ss_rr_W/X/Y/Z.SIM in the same folder). If the actual room model is not rectangular an approximate rectangular outline is created in the GEO-file for first order reflection generation in AniScape. An alternative is to use The FIReverb Suite (PureVerb) that also can be set to create AniScape files (Options menu)

227


5-channel (Receiver model)

Creates five different responses according to the 5-channel microphone setup selected. 5channel is mainly intended for use with the MultiVolver. _1, _2, _3, _4 and _5 are added to the ECH-file base-name corresponding to Center, Left, Right, Left surround and Right surround. 5-channel microphone setup (Receiver model)

The position, aim and directivity patterns for the five microphones. Pre-made types available are: Omni, Cardioid, Hyper-cardioid, Super-cardioid, Figure-of-8 and a special Figure-of-8+ (only the frontal lobe exists).

Load/Save

5-channel microphone setup

Loads and saves a 5-channel mic setup file (.CH5). CH5-files can be exchanged with PureVerb and are normally kept in CATTDATA-folder PV_MICS.

Plot

5-channel microphone setup

Plots the current setup and shows it in the stand-alone PLT-viewer. The plot includes the selected directivity patterns for reference and mouse-over location read-out is shown in the status-bar. Example shows L,C,R cardioids and Ls,Rs figure-of-8.

228


Output-files The resulting IR output data-files (Finite Impulse Responses, FIRs) will be saved to the same folder as the ECH-file and have names that are dependent on the receiver model selected.

The name syntax is (assuming that name.ECH was the original ECH-file name): Mono:

name_M.frm

Binaural/stereo:

name_L.frm (left) name_R.frm (right)

B-format:

name_W.frm (omni) name_X.frm (x-axis figure-of-8 ⇔ head-direction) name_Y.frm (y-axis figure-of-8) name_Z.frm (z-axis figure-of-8) nd rd for 2 order R, S, T, U, V and 3 order K, L, M, N, O, P, Q

5-channel:

name_1.frm (Center) name_2.frm (Left) name_3.frm (Right) name_4.frm (Left surround) name_5.frm (Right surround)

The extension .frm stands for format: MATLAB v4 binary

.MAT

MLSSA ASCII

.MLS

Lake DSP binary

.SIM

Windows WAV

.WAV (also has an additional _16 or _32 code added and _LR if stereo)

The SIM-file addition utility relies on the file-name syntax described above. See the Utilities menu for manipulating output-files. Binaural/stereo example names below for ECH-file E_A0_22.ECH.

229

7. POST-PROCESSING MODULE Lake SIM-format (.SIM) for the convolver (IR)

Output-files

Create a SIM-file for convolution using either the software convolver, MultiVolver or the Lake DSP convolution processors. This is a binary file in the Lake SIM-format. The format allows any number of comments before the file-header. The PP module uses this to document the prediction project name and some other parameters such as initial scale-factor needed for calibration. For Lake convolution the data is optimally scaled not to overflow the integer convolution. Software convolution auto-scales to give optimal dynamic range (introduces a second scale-factor). SIM-files will have the sample-rate and IR size selected in General settings. The header contents of a SIM-file can be inspected using a Utilities|Information submenu. Sample file-names: E_A0_22_L.SIM and E_A0_22_R.SIM. MLSSA ASCII-format (IR)

Output-files

Create an MLS-file for MLSSA [MLSSA]. This is an ASCII-file to be imported to MLSSA by the "TRANSFER IMPORT" command. An alternative to use this function is to use the file conversion utility from SIM to TIM format. The resulting file can then be loaded into MLSSA by the "TRANSFER LOAD" command. MLS-files will have the sample-rate and IR size as selected in General settings. Sample file-names: E_A0_22_L.MLS and E_A0_22_R.MLS. MATLAB binary-format (IR)

Output-files

Create a MAT-file for MATLAB v4 format [MATLAB]. This is a binary file. The BRIR vector (variable pp_ir) is calibrated i.e. no scaling is performed. The file contains various variables and is organized as follows (variable names): pp_initdelay pp_ir pp_intfs pp_auxdelay pp_procdelay pp_sincdelay pp_earlyp pp_latep pp_tailstart pp_lfsupp pp_supp pp_recmodel pp_micz pp_micx pp_decay pp_iacc

delay source to receiver [s] the impulse response (left or right calibrated) internal sample-rate [Hz] short delay caused by processing [s] electrical loudspeaker delay [s] overall delay in samples if fractional delay is used (else = 0) early transform-size late transform-size late part start [s] LF suppression on/off (1/0) suppression value, < 0 [dB] receiver model (0 = omni, 1 = binaural, 2 = B-format, 3 = mic, 4 = stereo, 5 = 5-channel) microphone parameter if pp_recmodel = 3 or 4 microphone parameter if pp_recmodel = 3 or 4 corresponding to selected late part decay, > 0 [dB] if the receiver model is binaural and IACC is selected a 9-element vector will give the calculated IACC values (elements 1..8 is 125..16k, element 9 is the overall value).

Sample file-names: E_A0_22_L.MAT and E_A0_22_R.MAT. 16-bit WAV uncalibrated (.WAV) (IR)

Output-files

This is a simple version most appropriate for a quick listen. I does not include the initial delay and is always scaled to 16-bit max. Mono, B-format and 5-channel IRs are saved as mono WAVs with the usual naming convention but with an added _16 (e.g. E_A0_01_16_W.WAV).

230


For convenience, stereo and binaural IRs are saved as stereo WAVs with the name syntax E_ss_rr_16_LR.WAV. 32-bit WAV calibrated (.WAV) (IR)

Output-files

This is a more advanced version most suitable for various kinds of independent further processing. The files include the initial delay (as well as any electrical aux delay for electroacoustic sources) and are always scaled in a consistent manner retaining individual IR levels. Since then the level span of a 32-bit file is so large (> 186 dB), a weak normal natural impulse response can use around 16 bits but PA IRs that may have much higher level due to the Gain_a values can utilize more bits and have the scale factor built in. It means that you have to use an editor that can read actual 32-bit PCM files and not one that just reads the upper 16 bits. Mono, B-format and 5-channel IRs are saved as mono WAVs with the usual naming convention but with an added _32 (e.g. E_A0_01_32_W.WAV). For convenience, stereo and binaural IRs are saved as stereo WAVs with the name syntax E_ss_rr_32_LR.WAV. (Plot-files)

Output-files

Which plot-files in CATT PLT-format to generate. A plot list-file POSTPLT.TXT is created in the folder where the ECH-file is read and the view module automatically displays any created files. Filenames below assume that the ECH-file was name.ECH. Impulse response (early part only)(Plot-files)

Output-files

Plot the IR with a length corresponding to 150% of the early part. Filename name_I.PLT. If nd rd B-format is selected any 2 order B-format components are included but 3 order is are plotted in a separate file name_I3.PLT. Octave-band decay curves (Plot-files)

Output-files

Plots of octave-band decays before IR synthesis. Least-square fitted lines and RT values are shown. The decays always correspond to an omni microphone and are only generated if at least -30 dB has been reached for all octave-bands (or decay curves and reverberation times would be meaningless). Filename name_D.PLT.

Early part response Creation of an early part of the IR based on the predicted early part echogram (corresponding to the upper right graph in the Full detailed calculation plot-file E_ss_rr_oct.PLT), see section 7.3. Exclude orders lower than

Early part response

This item is only available in a special version with addons for the Lake DSP AniScape application. For other software versions it is not displayed and the setting is always 0. It is a special purpose option to exclude e.g. the direct sound and all first order reflections (setting 2) to be handled in real time by some other method, e.g. by AniScape. Normal setting is 0 (i.e. direct sound and all early reflections are included). If not set to 0 a warning is issued if warnings are enabled.

231


Late part response Settings for creation of the late part of the IR (i.e. the reverberation tail) based on the a simplification of the hall model and statistically generated taking into account spatial incidence from the actual model, see section 7.3.

(IR truncation)

Late part response

At which level in the decay-curve relative the direct sound, to truncate the reverberation tail generation (for the band with longest RT). Default

Truncate the IR at the selected total IR size.

User defined

Selected by the user. A minimum of 40 dB is recommended (i.e. 2/3 of the RT) or the truncation may be heard. Be sure to match this value and the IR size so that truncation is done at the time intended (the decay can of course not be longer than the time corresponding to the selected IR size).

Create files for head-tracking This item is only enabled if binaural post-processing is selected and is useful either for including head rotations with the Walkthrough convolver or for the Lake DSP HeadScape application. Depending on application a different filename syntax is used for the created SIMfile, see Preferences. The processing needs a head-tracking data file with head rotation angles to be selected (.HTR). This file is for HeadScape use copied to the output folder after completed processing so when HeadScape is run all necessary files are present. Sample HTR-files are located in the CATTDATA-folder HRTF. This option creates one pair of long filters as usual but also a set of short early-part filters to be head-tracked using HeadScape or to be used in a Walkthrough convolver script. To run HeadScape the long filters and the HTR-file needs to be selected and the HeadScape application then expects to find the set of short filters in the same folder as the HTR-file. These files are preloaded into Huron RAM when the HeadScape Configure button is pressed. The tail filter will be truncated by HeadScape at 59392 taps. The structure of an HTR-file is: HURON1 N Hi [,] Vi ...

Where HURON1 file identification tag N length of the early head-tracked FIRs at the selected samplerate (≤ 3840 for HeadScape use).

232


Hi Vi

horizontal head angles from 0° to 360° (0 is straight ahead and > 0 is head turned to the left). vertical head angles in from –30° to 90° (0 is straight ahead, < 0 is looking up and > 0 is looking down).

A maximum of 128 angle pairs (for HeadScape use) can be specified in any order and angle values can be specified with an accuracy of 0.5°. Decimal point must be used. For HeadScape use the file-name syntax is : HhhhvvvL.SIM and HhhhhvvvR.SIM.

For Walkthrough convolver use the syntax is: E_ss_rr_hhhvvv_L.SIM and E_ss_rr_hhhvvv_R.SIM,

the first part of the syntax is thus the same as for the long filter. Where hhh vvv

is the horizontal angle in the number of half degrees calculated as 2⋅Hi giving values in the range 0..719. is the vertical angle in the number of half-degrees calculated as 2⋅(90 + Vi)giving values in the range 120..360.

Example from E_A0_12.ECH: Files created for HeadScape use: HURON1 3984 0.0 0 0.5 0 5.0 5.0 5.0-5.0 ...

H000180L.SIM H001180L.SIM H010190L.SIM H010170L.SIM

H000180R.SIM H001180R.SIM H010190R.SIM H010170R.SIM

Files created for Walkthrough use: HURON1 3984 0.0 0 0.5 0 5.0 5.0 5.0-5.0 ...

E_A0_12_000180_L.SIM E_A0_12_001180_L.SIM E_A0_12_010190_L.SIM E_A0_12_010170_L.SIM

E_A0_12_000180_R.SIM E_A0_12_001180_R.SIM E_A0_12_010190_R.SIM E_A0_12_010170_R.SIM

For HeadScape use details, and for changes in possible filter lengths, consult current HeadScape documentation.

Create CWI-file for CATT-Walker This item is only enabled if B-format post-processing is selected in General settings (with 1st order and AniScape off). Via this special post-processing real time walkthrough auralization and visualization is available using the standalone external software CATT-Walker (Walker.EXE, see File|External CATT Tool). An overview of the procedure, and how to operate CATT-Walker, can be found in Walker.PDF.

233


Post-processing for CATT-Walker is similar to normal post-processing and works as follows: st

·

select General settings as normal and Receiver model 1 order B-format and AniScape off

·

select Create CWI-file for CATT-Walker on the main dialog (enabled when B-format is on)

·

Browse for a CAG-file (as was created in the output file along with the ECH-files for all receiver positions). When the CAG-file is read, and if it has more than one source (if the prediction was made using several sources), a dialog is shown where the source to use can be selected.

·

press Save and Run as usual and all receivers defined will be post-processed (note a count-down and a timer on the Processing dialog estimating the total remaining time). When finished a single file called WALKER_ss.CWI is created where ss is the id of the source used (CWI = CATT-Walker Impulse responses, see Utilities|Information|CATT CWI-file for info on a created CWI-file).

·

start CATT-Walker via File|External CATT Tool

·

press Help and read the dedicated Walker.PDF help-file, keep this file open for reference.

·

press Config and select the CAG-file, the corresponding CWI-file and the desired anechoic WAV-file.

·

press Start, for operation details see Walker.PDF.

Since a PST-file includes the settings also for CATT-Walker post-processing it can be used in the Sequence processing module. Since post-processing to create a CWI-file for CATTWalker does not require an ECH-file to be given, (NONE) can be given in the ECH-file field in the Post-processing sequence step. CWI-file IRs can be analyzed via Utilities|Information|CATT CWI-file. To add several CWI-files together for a multi-source PA system walkthrough (assuming the same anechoic input to all sources) use Utilities|CWI-file Addition.

Convolution Utility to run software convolutions. Convolves one- or two-channel impulse responses in Lake DSP SIM-format with anechoically recorded mono material in WAV-format. For convolution requiring > 2 channels the MultiVolver can be used. Floating-point convolution is used and any length of FIR can be convolved with any length of input-material (some temporary hard-disk workspace is required, see below). The resulting, processed, file is in WAV-format autoscaled to 16 bits for maximum dynamic range. A binary file containing scale-factor used and some other data is also created, the filename is Name.SCL if Name.WAV is the selected output filename, see below. The SCL-file contains all necessary information for the WAV-file calibration and addition utilities to perform relative calibration of processed files. To enable WAV-file calibration, and addition, a processed WAV-file needs to be accompanied by its SCL-file whenever it is copied to another location. The file info and WAV-file scaling/calibration/addition utilities reads the SCL-file if present. For details of calibration see 7.3 Theory. The sample-rate may be defined in the SIM-file header. If it is, it is checked so that it matches the sample-rate of the selected anechoic WAV-file. Note: The convolver may need temporary disk workspace of at least nine times the size of the input WAV-file or the output file will be truncated or may contain rubbish data. There are two reasons to this. The first reason is that the convolver uses a caching technique so that the first

234


time an anechoic file is processed for a particular case it takes slightly longer but any subsequent convolutions will run twice as fast. The cache information is stored in PPW-files in the CATTDATA-folder ANECHOIC and can be deleted at any time since they will be created again if needed (they are quite big for long WAV-files). The second reason is that the processing, if the anechoic WAV-file is large (> ca. 20 sec at 44.1 kHz), creates a temporary floating-point file that then is scaled to a full 16-bit WAV-file, these file are deleted after completion. It is recommended that all anechoic mono WAV-files follow a certain filename convention and they must be placed in the CATTDATA-folder ANECHOIC. The naming convention is: mm_ff_AN.WAV where mm is the name of the source material (for example FT for Female Talker), and ff is the sample-rate (44 for 44100 Hz), and AN is for anechoic. A full example is thus FT_44_AN.WAV. If this file is convolved with E_A0_01_L.SIM and E_A0_01_R.SIM the processed file could be named FT_A0_01.WAV. This way both the source material (FT) and the source/receiver combination (A0_01) can be tracked. The sample-rate is stored in the WAV-file header and can be checked using the Utilities| Information menu along with other info. Note: the anechoic material provided with the full software version is copyrighted, see Section 1.1. For limited versions only a single 22050 Hz file can be used.

Name (Input (anechoic) WAV-file)

Convolution

Anechoically recorded mono WAV-file from the CATTDATA-folder ANECHOIC. The default file can be specified in Preferences. If non-anechoic material is used (e.g. binaurally recorded material that is to be cross-talk canceled it must also be put in ANECHOIC). Mono

the specified input WAV is mono

Stereo left, Stereo right the specified input WAV is stereo and the left or the right channel is to be used as mono input (typically for cross-talk cancellation of binaural recordings or auralization of stereo speakers). Stereo mix the specified input WAV is stereo and the sum of the left and the right channel divided by 2 is to be used as mono input

235

7. POST-PROCESSING MODULE 2-channel convolution (binaural or stereo) (Impulse responses (SIM-files))

Convolution

If two-channel (stereo or binaural) or one-channel (mono) convolution will be performed. Play (Impulse responses (SIM-files))

Convolution

Plays the selected IR(s) over the soundcard. Left (Impulse responses (SIM-files))

Convolution

Selection of the left (or mono) filter SIM-file. If not mono and if the file is the left channel of a stereo IR (E_ss_rr_L.SIM), the right IR SIM is automatically filled in. Right (Impulse responses (SIM-files))

Convolution

Selection of the right filter SIM-file (if 2-channel). Name (Processed WAV-file (to be created))

Convolution

Desired name of the processed file. See recommended file-name syntax above. Auto play after completed convolution (Processed WAV-file (to be created))

Convolution

Launches the CATT WAV-file player and automatically plays the file. Use less harddisk space

Convolution

Does not use the cashing technique with PPW files as discussed above. Could be used to save temporary harddisk space. When convolution is selected as a Sequence module step the variant that is faster for repeated similar cases is always used since that is a typical use of a sequence (to convolve e.g. responses from 16 positions in a room with the same WAV). For very long WAV-files the MultiVolver can be used since it does not use any extra hard disk space at all. Also, the WAV-files to be processed do not have to be in the ANECHOIC folder. On the other hand the MultiVolver does not create any SCL-files needed for calibration and WAV-file addition. OK

Convolution

Performs the convolution and optionally plays the file when finished.

Walkthrough convolution To be able to hear the sound of moving around in a room, the walkthrough convolver can be used. By calculating room responses along a walking path and associate each with a time the convolver interpolates and creates the effect of walking along the path while listening to the same sound playing. This convolver can also be used to create smooth transitions (morphing) between cases in e.g. continuos A/B tests (e.g. with or without room treatment, different loudspeaker alternatives etc). Note: Doppler is not included. The external CATT-Walker (see File|External CATT Tool) will perform a similar task but in real time while moving around in the virtual room, the Walkthrough convolution utility is better suited for creating auralized sound tracks to rendered walkthrough videos.

236


Input (anechoic) mono WAV-file

Walkthrough convolver

Anechoically recorded mono WAV-file from the CATTDATA-folder ANECHOIC. The default file can be specified in File|Preferences. If a too short input WAV-file is chosen, it will automatically be looped so that it is not "exhausted" until the list of filters has been run through. However, it may sometimes anyway be better to manually create a long enough looping sound in a WAV-editor so that there is a good loop rhythm. (Walkthrough Convolver Script-file)


A script-file that defines how the convolution is to be performed (.WCS). The syntax of a WCSfile is (note the Create button described below): CATT-WCS VERSION = version SIMFOLDER = "simfolder" INCHANNELS = inchannels INCHANNELSUFFIX = insuffix OUTCHANNELASSIGN = outassign MODE = mode UNIT = unit time "basename" ["tailname"]

... Where: version

WCS-file version, 0 for v8.0

simfolder

folder where all SIM-files are located, with full path given in ""

inchannels

number of input channels (e.g. 2 for binaural/stereo, 4 for B-format, 5 for 5channel)

insuffix

the SIM-filename characters identifying the channel (e.g. "_L" "_R" for binaural or stereo). This suffix will be added to each SIM-file basename in the list, followed by ".SIM"

237

7. POST-PROCESSING MODULE outassign

mono|stereo , i.e. if one mono WAV per channel is to be created or instead a set of stereo files. For binaural stereo is the normal choice but for B-format and 5-channel mono is often most practical. If an odd number of channels are used with stereo assign (e.g. WXY), the last stereo file created with have an empty right channel.

mode

time_relative|time_absolute i.e. if the time associated with each response is relative or absolute

unit

ms|s|frames fps|fractions f the time unit time associated with each response given: milliseconds, seconds, frames with the given framerate (fps frames per second), or sec fractions with the given fraction f.

time

the time during which the given response is used (relative or absolute according to header MODE)

basename

is the base name of the SIM-file given in "", e.g. "E_A0_12"

tailname

optional: if basename is the SIM-file for head-tracking (e.g. "E_A1_23_000180") tailname is the full response for the corresponding position, i.e. "E_A1_23". The tail FIR is first loaded and then the first part is repeatedly overwritten by the early part FIR before convolving.

Example of a script where the walkthrough goes from position 13 to 01, turns around, and walks back again looking backwards: CATT-WCS VERSION = 0 FOLDER = "c:\v8doc\" INCHANNELS = 2 INCHANNELSUFFIX = "_L" "_R" OUTCHANNELASSIGN = stereo MODE = time_relative UNIT = frames 30 24 "E_A0_13" 24 "E_A0_12" Walking from 24 "E_A0_11" 13,12,11…03,02,01 ... 10-04 removed here 24 "E_A0_03" 24 "E_A0_02" 96 "E_A0_01" 4 "E_A0_01_000180" "E_A0_01" 4 "E_A0_01_010180" "E_A0_01" Turning around using early part 4 "E_A0_01_020180" "E_A0_01" HeadScape files ... 030180-330180 removed here E A0 01 000180 360180 4 "E_A0_01_340180" "E_A0_01" 4 "E_A0_01_350180" "E_A0_01" 4 "E_A0_01_360180" "E_A0_01" 96 "E_A0_01_360180" "E_A0_01" 24 "E_A0_02_360180" "E_A0_02" 24 "E_A0_03_360180" "E_A0_03" Walking from 01,02,03…11,12,13 24 "E_A0_04_360180" "E_A0_04" (cheating a bit using the old tail and just rotate the ... 05-10 removed here early part using HeadScape files) 24 "E_A0_11_360180" "E_A0_11" 24 "E_A0_12_360180" "E_A0_12" 24 "E_A0_13_360180" "E_A0_13"

The external CATT tool WalkDecim (see File|External CATT Tool) can greatly simplify the task of creating WCS files since it can import a camera path from a rendering program and

238


from that create a receiver file as well as a corresponding WCS. Create (Walkthrough Convolver Script-file)


A utility dialog that creates a basic script-file that can be filled in with a list of times and SIMfiles. Edit items correspond to WCS-file header items described above.

Processed WAV-file base-name, to be created


The resulting WAV-file base-name. The full name is created by adding the given INCHANNELSUFFIX and ".WAV". Examples: Binaural/stereo out-assign (one stereo file): basename_L_R.WAV. Binaural/stereo and mono out-assign (two mono files): basename_L.WAV and basename_R.WAV. B-format and mono out-assign (four mono files): basename_W.WAV basename_X.WAV basename_Y.WAV basename_Z.WAV. 5-channel and mono out-assign (five mono files): basename_1.WAV basename_2.WAV basename_3.WAV basename_4.WAV basename_5.WAV.

239

7. POST-PROCESSING MODULE Use shorter proc. frames (not required for speech/music)

Normally this option is not necessary with speech and music but can be useful with special test signals like pink noise or pure tones that are quite sensitive and where the impulse response interpolation can be audible. The shorter frame option causes the processing to take longer time.

Convolution (WAV impulse responses) Utility to run software convolutions from e.g. measure impulse responses. Convolves one- or two-channel impulse responses in WAV-format with anechoically recorded mono material in WAV-format. For convolution requiring > 2 channels the MultiVolver can be used. Floatingpoint convolution is used and any length of FIR can be convolved with any length of inputmaterial (some temporary hard-disk workspace may be required, see below). The resulting, processed, file is in WAV-format auto-scaled to use all bits of the selected format for maximum dynamic range. Note: The convolver may need temporary disk workspace. If the anechoic WAV-file is large (> ca. 20 sec at 44.1 kHz), the processing creates a temporary floating-point file that then is scaled to a bit-depth WAV-file, this file is deleted after completion.

The free GratisVolver performs similar functions and can also play selected WAVs. Name (Input (anechoic) WAV-file)

Convolution (WAV IR)

Anechoically recorded mono or stereo WAV-files from the CATTDATA-folder ANECHOIC or free location (see Any Folder). The default file for normal SIM-file convolution is pre-selected (can be specified in Preferences). Mono

the specified input WAV is mono

Stereo L

the specified input WAV is stereo and the right channel is to be used as mono input.

Stereo R

the specified input WAV is stereo and the left channel is to be used as mono input.

240


Stereo mix the specified input WAV is stereo and the sum of the left and the right channel divided by 2 is to be used as mono input

The input WAV file can be 16, 24 or 32-bit PCM WAV. Any folder (Input (anechoic) WAV-file)


Selection of the input WAV-file can be from any hard disk location. Name (Mono or stereo impulse response WAV-file)


Selection of the impulse response WAV-file. The file can be can be 16, 24 or 32-bit PCM WAV. If the file is mono, 1x1 mono-to-mono convolution is made and a mono output WAV is created. If the WAV is stereo, 1x2 mono-to-stereo convolution is made and a stereo output WAV is created. Play (Mono or stereo impulse response WAV-file)


Plays the selected IR over the soundcard. Note: this is meant for short WAVs like IRs since it can't be stopped until playing is over, to play longer files use Utilities|WAV-file Player. Auto play after… (Processed WAV-file (to be created))


Launches the CATT WAV-file player and automatically plays the file. Bits (Processed WAV-file (to be created))


The created WAV-file can be 16, 24 or 32-bit PCM WAV or take the bit depth of the input, if one of these. Name (Processed WAV-file (to be created))


Desired name of the processed file. For very long WAV-files MultiVolver can be used since it does not use any extra hard disk space at all.

241


7.3 Theory This section describes the background to the IR synthesis performed by the PP module with some references to literature.

Impulse response scale and calibration Created Impulse Responses (IRs) are normalized by not including the acoustical input spectrum (Lp1m_a in the source file) since the it is replaced by the spectrum of the anechoic file at convolution. By removing the acoustical source spectrum it is also reasonable to switch anechoic material as long as the implied source has a directivity pattern similar to the one given in the source-file. The finally heard processed music/speech has been scaled in several steps, see Fig. 7.2. Prediction Post-processing (sp) Performance

xPa

h sp Digital recording Convolution (sc) at 1 m distance (sa) s x y a Pa

scorr

Playback system

yPa

Fig. 7.2 principal scale factors from anechoic recording to audible auralization.

(

)

y = ( x Pa s a ) h s p s c

(7.3-1)

Where y

final processed music/speech 16-bit integer format

xPa true pressure signal of anechoic music/speech material at 1 m in front of the real source [Pa]. xPa includes an extra, typically unknown, scale-factor if the recording was made at another distance. -1

sa

scale-factor when xPa was digitally recorded to 16-bit integer format [Pa ]. For the majority of available anechoic material this factor is unknown.

h

impulse response as described above

sp

scale-factor applied at the post-processing in the PP module. The reason for this factor is the 24-bit integer SIM-file format used by the Lake DSP convolvers. The factor is saved in the SIM-format output-files header created by the PP module. The SIM-file file information utility displays the value and the calibration and scaling utilities updates it.

sc

scale-factor applied at convolution to create 16-bit integer format. When using Lake DSP hardware this factor is unity. When software convolution is used this value is stored in the SCL-file with the same base-name as the processed material. The WAVfile information utility displays this value.

242


To arrive at absolute calibration the total scaling is thus:

(

y Pa = ys corr ; s corr = s a s p s c

)

−1

(7.3-2)

Where yPa is the true pressure signal at the receiver position used in the particular hall modeled [Pa]. The playback system must then create this signal at the ears of a listener. scorr is the correction scale-factor [Pa]

As can be seen from the various scale-factors above, absolute calibration may be very difficult to achieve and requires calibrated amplifiers and well known anechoic source material. Presently, the head-phone equalization filters applied for a binaural receiver are also not calibrated in level. Relative calibration for a particular anechoic source signal is, however, achievable. The relative level variations within a hall (or across halls) can be preserved by observing the two factors sp and sc. Since the software convolution auto-scales for maximum dynamic range the calibration for software convolution must be done after convolution by observing spsc. The WAV-file calibration and addition utilities do this automatically and, by using the corresponding SCLfiles, allows only for calibration of processed files that have exactly the same anechoic sourcefile. When using Lake DSP hardware the calibration must be done of the IRs before convolution so that the scaling of all responses are equal. See the post-processing flow illustration in Fig. 7.1 for an overview of both desktop/software and Lake DSP auralization including relative calibration.

Reflection path (RP) transfer function The prediction module calculates early part octave-band echograms (125 Hz to 16 kHz). These echograms are saved in the ECH-file together with air absorption, reverberation time in octave-bands and other information.

Reflection path transfer function magnitude Depending on reflection order, two different methods for reflection path magnitude synthesis methods are employed. For the direct sound (if present) and first-order specular reflections

Frequency lines in-between the octave-band center-frequencies are interpolated using a cubic spline. To assure a smooth interpolation it is made for log(magnitude) vs. log(frequency). The interpolated function is then returned to lin(magnitude) vs. lin(frequency) and sampled at the requested number of points (half of reflection transform-size). The part below the lower limit of the 125 Hz octave-band (88 Hz) can be treated in two ways: either suppression by a raised cosine bell or a flat response down to 0 Hz, see Fig. 7.3.

243

7. POST-PROCESSING MODULE For first-order diffuse reflection and all second- and higher order reflections

Reflections are octave-band based since they originate from independent RTC tracing of each octave-band and cannot be splined. The reflections are instead built up by digital octave-band filters that are added together.

Fig. 7.3 Reflection path magnitude spline interpolation

Reflection path transfer function phase Depending on reflection order, two different methods for reflection path phase synthesis methods are employed. For the direct sound (if present) and first-order specular reflections

A minimum phase is constructed from the magnitude using Hilbert transform relations [Oppenheimer]: arg[H(f )] = H (log | H(f ) |) Where H(f ) arg[H(f )] |H(f )|

H

(7.3-3)

RP transfer function RP transfer function phase (without the frequency independent linear phase shift corresponding to the RP total delay which is accounted for separately) RP transfer function magnitude Hilbert transform

For first-order diffuse reflection and all second- and higher order reflections

For these reflections details are not important and the phase becomes that of the linear phase octave-band FIRs used (the filter delay is compensated for).

244


Receiver model From the previous two sections the complete RP transfer function from the (point) source to the (point) receiver is estimated.

Binaural For a binaural room IR synthesis the next step is to calculate the pressure at the two ears of an imaginary head thereby splitting the point RP transfer function into left and right ear RP transfer functions which include the phase difference between the ears and the shadowing effect of the head. This is done using head-related transfer functions (HRTFs). The initial v7 HRTFs are measured at 15° elevation steps. In azimuth, measurement points are 15° apart from 30° below to 30° above the horizontal plane (i.e. for elevation -30°, -15°, 0°, 15°, 30°). For higher elevation angles (45°, 60°, 75°) the azimuth steps are 20°, 30° and 60° respectively). Impulses arriving with angles of incidence more than 30° below the horizontal plane are assigned to the lowest sector, see Fig. 7.4.

Fig. 7.4 HRTF measurement positions

For the early part synthesis horizontal (azimuth) HRTFs are interpolated but in the vertical (elevation) direction the closest measured angle is used. For ≥ v8.0c HRTFs can have any resolution and full 3D interpolation is performed between measured points. An example of the newer resolution is the ITA1_plain_44.DAT and ITA1_plain_48.DAT libraries that are based on the same measurement as the previous ITA0 but uses the original resolution of 5.6° horizontal and 15° vertical. The penalty of the more general HRTF-handling is a slightly slower early part binaural post-processing. For the late part the space surrounding the receiver is divided into seven sectors:

245


1) 2) 3) 4) 5) 6) 7)

above the head in front of the head to the left and in front of the head to the right and in front of the head to the left and behind the head to the right and behind the head below the head (derived from an average of the lowest ring HRTFs)

At Full detailed (RTC) prediction the incidence distribution according to these 7 sectors is recorded and applied when the tail is created. For each sector and late part reflection the HRTF is selected randomly from three positions within the sector, see Fig. 7.4 and 7.5. right back

right side

6

4

left back

5 top 1

3

left side

2 front

Fig. 7.5 HRTF tail sectors for sector processing, bottom not shown

The final BRIR is equalized for use with headphones depending on the settings in File|Preferences. For further details about the headphone eq. and the HRTFs (e.g. to test or create further HRTF libraries) check the CATT users’ www page (See Help|About for a link).

Mono and Blumlein/XY-stereo For mono microphone or stereo IRs, angle dependent compensation is made according to the directivity selected. Microphones are ideal as were they of infinitesimal size and the directivity pattern is implemented as F(θ ) = Z + X·cos(θ ) where θ is the angle of incidence relative the main axis of the microphone and where Z+X = 1 (Z and X ≥ 0). For details see [Borwick]. With mono the microphones are aimed according to the head direction selected at prediction, with stereo the stereo pair is aimed according to the head direction.

B-format st

Similarly for the 1 order B-format WXYZ response, microphones are ideal and perfectly coincident where figure-of-8 types are pointing in the X, Y, and Z directions. According to the most common B-format specification X is pointing in the head-direction, Y to the left, Z up. The nd omni response (W) is by convention scaled down by a factor 1.414. With the 2 order rd (RSTUV) and 3 order (KLMNOPQ) options the encoding is as follows: R

3*(Z*Z - 1)

S

2*Z*X

T

2*Y*Z

U

X*X - Y*Y

V

2*X*Y

K

Z*(5*Z*Z-3)/2

246


L

8*X(5*Z*Z-1)/11

M

8*Y(5*Z*Z-1)/11

N

Z*(X*X-Y*Y)/2

O

X*Y*Z

P

X*(X*X-3*Y*Y)

Q

Y*(3*X*X-Y*Y) st

where X, Y and Z are the 1 order components of an individual reflection. SIM-files created will st have _R, _S etc. added just like 1 order _W, _X etc. Note that PureVerb v3.2e SIM to WAV conversion (File menu) will convert also the RSTUV and KLMNOPQ files if found for those nd rd that want to use 2 and 3 order tools with WAV-format filters (no CATT tools except SIM-file nd rd addition currently handles 2 or 3 order).

5-channel 5-channel is similar to B-format but microphones are typically not coincident and are of arbitrary types. Common to all 5-channel setups is that they are left/right symmetric, the center microphone (1/C) is pointed forward, the left microphone (2/L) is pointed forward or to the left and that the left surround microphone (4/Ls) is pointed backwards (or backwards to the left) and that the surround mics are placed behind LCR. The microphone configuration affects the sound considerably and there are several different "schools". Therefore, 5-channel auralization is not recommended for normal auralization since the microphone setup used affects the perceived soundfield considerably but may have its uses also there (e.g. for more informal presentations to a group of listeners). It also offers an opportunity to use the MultiVolver for 5-channel responses created for models of actual rooms, such as e.g. well sounding churches or concert halls, to be used for surround sound music reverberation purposes.

Late part generation The late part reverberation tail is based on statistical reflection generation in a simplified version of the hall model where natural distributions or reflection path lengths and absorption factors are sampled. For faster processing, the number of reflections are decimated (sector processing, see Fig. 7.5) by first creating sector echograms. First the intensity from each virtual reflection in each octave-band is added into seven sector echograms with 1 ms slotsize. Then a reflection is synthesized for each slot in each sector. Sector processing is a simplification based on limitations in the human auditory system and is much faster than detailed processing of each single reflection. The result is difficult to distinguish from detailed processing. Since there are seven late-part sectors for binaural processing, a 1 ms slot-size can give 7000 reflections/s in the late part. Incidence angles are sampled from actual sevensector incidence angle distributions recorded during prediction. This approach will approximately preserve spatial reflection distribution and reflection density at the receiver. The PP module optionally generates a plot-file (name_D.PLT in the same directory as the ECH-file as selected in the Output-files dialog) showing the resulting echogram decays for octave-bands 125 Hz to 16 kHz. The conversion of octave-band power-based echograms to impulse responses inevitably introduces some types of difference in the energy parameters such as T-30 and D-50. The change is most noticeable at 125 Hz and gradually decreases up to 1 kHz where the addition of reflections with synthesized phase in practice will give the same result as the addition of

247


power-based reflections in an echogram. From v8.0b a new procedure makes the late energy of a synthesized impulse response correspond better to that of the prediction module at low frequencies. However, it must be stressed that even if the correspondence between auralized and predicted intelligibility is increased it does not mean that the prediction itself at low frequencies is improved since when using rays and geometrical acoustics, the predictions at low frequencies are not as good as that at mid and high frequencies so in all cases it is wise to apply caution and never over-interpret the results at low frequencies.

Inter-Aural Cross Correlation Coefficient IACC, the Inter-Aural Cross Correlation Coefficient, gives an estimate on the similarity between the left and the right ear and is defined as follows [ISO][Cremer]. First, the Inter-Aural Cross Correlation Function:

t 2  ( τ ) = ( t ) ( t + τ ) dt IACF t 1t 2  ∫ hl hr  t1 

t2 t 2 2  2 h h ( ) t dt ( t ) dt r ∫ ∫ l  t1 t 1 

−1/ 2

(7.3-4)

The Inter-Aural Cross Correlation Coefficient:

IACC t 1t 2 = max IACF t 1t 2 (τ ) for τ min < τ < τ max

(7.3-5)

Where hl hr t1 t2

τmin τmax

left ear response right ear reponse lower integration time limit, 0 ms upper integration time limit, 80 ms lower time limit for evaluation, -1 ms upper time limit for evaluation, +1ms

IACC is evaluated in octave-bands 125 Hz to 16 kHz. IACC can be calculated from a binaural SIM-file pair and is selected from the Utilities menu or be requested to be calculated at processing and put in the SIM-file header, see General settings.

Crosstalk canceling filters Crosstalk canceling enables loudspeaker listening to binaurally recorded or simulated sound. The technique is sometimes called "transaural". The background is that when listening to a binaural recording over loudspeakers (in a typical symmetrical stereo configuration), the left as well as the right ear is reached by sound from both left and right speakers while for binaural replay the left signal is supposed to be sent only to the left ear and the right signal only to the right ear [Cooper-89], [Kotorynski-90].

Filter structure The implementation of crosstalk cancellation is based on the "Shuffler" filter configuration:

248

7. POST-PROCESSING MODULE L

R

+

Binaural without headphone eq

-

1 S+A

1 S-A

-

A A

S

S

+

L’

R’

For loudspeakers

Fig. 7.6 Filter structure for crosstalk cancellation, definition of S and A.

where S

is the HRTF from the left speaker to the left ear as well as (because symmetry is assumed) from the right speaker to the right ear (S = Same side).

A

is the HRTF from the left speaker to the right ear as well as (because symmetry is assumed) from the right speaker to the left ear (A = Alternate side).

With the implementation, S and A are taken from the HRTF library currently selected in Preferences and interpolation is used to find the HRTFs for the exact angles used in the specified loudspeaker setup. When dedicated crosstalk cancellation filters are created their operation are according to Fig. 7.7 where Xt was the name given in the Crosstalk cancelling dialog. To use the setup with a Lake DSP Huron, a 4-channel convolver has to be set up and the resulting sounds must be added according to the figure. With software convolution it is easiest to use the MultiVolver with a 2x2 setup. L

Xt_L_L.SIM

Binaural recording

Xt_R_L.SIM

Xt_L_R.SIM

+ L’

R

Xt_R_R.SIM

+ For loudspeakers

R’

Fig. 7.7 Crosstalk cancellation of a binaural recording

Filter processing The processing involves inverting the two filters F1=S+A and F2=S-A. Had both filters been minimum phase except for a pure delay it would have been straightforward but that is seldom the case. While creating the filters it is possible to simulate the listening and see how flat the direct signals are (L to L) and (R to R) and how much the crosstalk (L to R and R to L) is suppressed. With the exact version of the filters (Smoothing=None for both HRTFs and filters

249

7. POST-PROCESSING MODULE F1 and F2) and with the DSP applied, the theoretical crosstalk cancellation is well sufficient (typically >> 40 dB) and the direct signals are perfectly flat (typically within 0.1 dB). This all assumes: 1) perfect and equal loudspeakers 2) no influence from the room 3) the listener is the one from which the S and A HRTFs have been measured 4) the listener sits exactly in the "sweet-spot" looking straight ahead Let us examine these four points in order: 1) even high quality loudspeakers are not perfectly equal in magnitude and phase response. The implementation has an option for individual loudspeaker correction. 2) wall reflections will be mixed with the created signals and degrades the crosstalk cancellation ability. Correcting the room is a much tougher task than correcting the loudspeakers. 3) this may happen but typically most users do not have their own HRTFs measured so that they can be used for the S and A filters. The implementation interpolates from the horizontal HRTFs to get the HRTFs for the exact angles to the loudspeakers. 4) since a static (non-adaptive) cancellation technique is applied the head must be kept inside fractions of a wavelength (at 10kHz the wavelength is about 3 cm...). The implementation has an option for low-pass filtering at 10 kHz to give a wider sweet-spot. Points 3) and 4) clearly suggest that one should not use too exact filters in the implementation. If a perfect filter is used there is a risk of audible high-frequency coloration if the head is moved out of the sweet.spot. The implementation has several options for individual smoothing of both the HRTFs used and of F1 and F2 (1/12-, 1/6-, 1/3 and 1/1-octave). Typically, 1/1octave smoothing is likely to always be too coarse and should be used only for experiments. The smoothing is applied as follows: HRTFs smoothing (optional): a) a pure delay between S and A is estimated (A is always delayed in relation to S but it is typically not a pure delay) b) the S and A magnitudes are smoothed using the selected 1/n-octave filter. c) minimum phase functions are created for S and A, and A is then delayed according to a) d) the new S and A are treated as if they were the actual S and A. Filter smoothing (optional): a) F1=S+A and F2=S-A are created and their magnitudes are smoothed using the selected 1/n-octave filter. b) inversion of F1 and F2 (not all implementation details are disclosed for this step)

Listening experience, discussion A successful application of a crosstalk cancellation system, as described above, depends on many factors. Below are some discussion points:

250

7. POST-PROCESSING MODULE 1) if the HRTFs used for the simulation do not give a good localization using headphones, transaural will not make it better. 2) generally, it is much easier to perceive the sounds as coming from outside the head than when using headphones. It is often easier to appreciate for someone not used to listening in headphones. 3) the problem of headphone equalization is removed, but instead a problem of loudspeaker and room correction appears. 4) the loudspeakers can be corrected for fairly well while the room is much more difficult to correct for. The loudspeaker responses (to be inverted and used as correction filters) should ideally be measured at the listening position (because of cabinet diffraction effects). When measuring loudspeakers it is best to average over a head-sized volume around the sweet-spot in an anechoic room (or large enough room if MLS or similar is used) and to smooth the high frequencies. Also don't try to correct for too low frequencies or the loudspeakers may suffer. 5) basic requirements on the room are:

⋅

·

as absorbent as possible and with a non-colored sound;

·

symmetrical room with symmetrical loudspeaker placements relative the room

·

either soffit-mounted loudspeakers or speakers placed not too close to the walls. Nevertheless, good results may sometimes be achieved in normal rooms e.g. for demo purposes. Even if the crosstalk cancellation technique is associated with some problems it is very well worth exploring!

251


252

8. SEQUENCE PROCESSING MODULE

8. SEQUENCE PROCESSING MODULE This module can handle and perform a sequence of processing steps such as the complete sequence from prediction over post-processing, convolution to WAV-file calibration.

The module uses only the File menu with obvious actions on SEQ-files. File|Insert Sequence |Before/After Current Step will open an existing sequence and insert its contents before or after the currently selected step in the list. This will e.g. enable having one sequence for prediction, one for post-processing and one for convolution and then eventually make a complete prediction -> post-processing -> convolution sequence by inserting the three lists into a new list. When Save and Run from top is pressed the current sequence-file is saved and the steps are performed in sequence. If Save and Run from current is pressed the processing instead starts from the currently selected step (note that any defined variables will be known also if they are defined before the currently selected step). Before the processing starts, the Prediction and Post-processing module dialogs are automatically closed and any non-saved settings-files are prompted for to be saved. If warnings are on, they are ignored or it may halt the sequence waiting for a response. No plotfiles will be shown automatically by the view-module, but have to be viewed after the sequence has completed one by one using the View module's File|Open or the complete plot-file list using List|Open and selecting the appropriate PLT_FULL.TXT (e.g.) or POSTPLT.TXT plotfile lists. To view finished plot-files before a sequence is completed, use the stand-alone PLTfile viewer. The error-handling options (ignore, prompt, abort) are to be set according to the dependencies in the sequence. For example, if step 1 is prediction and step 2 is post-processing of the results from step 1, an error in the prediction should abort the sequence, while an error in post-processing of one ECH-file does not effect a subsequent post-processing or prediction. If a running step is aborted (using the Processing dialog Stop button) a prompt is displayed with an option to continue or abort the sequence.

253

8. SEQUENCE PROCESSING MODULE When reasonable, file-names are checked for existence (e.g. prediction PRD-files, postprocessing PST-files, and anechoic WAV-files) but in many cases a file used in a later step may have been created in a previous step (e.g. ECH-files or SIM-files). In such cases only the folder is checked for existence. Add VARIABLE... Displays a dialog where string variables can be defined (variables have no use for error actions). A new entry in the list is created.

String variables can be used to create generic sequences that can be re-used. An example (here shown as text): VARIABLE -i PATH | c:\halls\theater1 Prediction -a \calc.PRD Post-processing -i \E_A0_01.ECH \catt.PST VARIABLE -i PATH | c:\halls\theater2 Prediction -a \calc.PRD VARIABLE -i SS | A0 Post-processing -i \E__01.ECH \catt.PST Post-processing -i \E__02.ECH \catt.PST Post-processing -i \E__03.ECH \catt.PST

From the above can be seen that when a variable is used it is added like and a variable can also be given a new value in a later step. The actual variable strings are not resolved until the sequence is run. Strings and variable names can have up to 256 characters and use all but the < > characters.

254


Add REPEAT... Displays a dialog where a repeat step can be defined. A new entry in the list is created.

The Variable name will be expanded to two characters if the name length is 2 (e.g. RR typically for receivers ids : 01, 02, 03 etc.) but to one character if the name length is 1 (e.g. S typically used with source ids A : A0, A1, A2 etc). An example: REPEAT 8 RR 10 1 Post-processing C:\TEMP\E_A0_.ECH C:\bin.PST

Will perform the same as: Post-processing Post-processing Post-processing Post-processing Post-processing Post-processing Post-processing Post-processing

C:\TEMP\E_A0_10.ECH C:\TEMP\E_A0_11.ECH C:\TEMP\E_A0_12.ECH C:\TEMP\E_A0_13.ECH C:\TEMP\E_A0_14.ECH C:\TEMP\E_A0_15.ECH C:\TEMP\E_A0_16.ECH C:\TEMP\E_A0_17.ECH

C:\bin.PST C:\bin.PST C:\bin.PST C:\bin.PST C:\bin.PST C:\bin.PST C:\bin.PST C:\bin.PST

For general counting up to 999 using a length of 3 will expand to 000, 001,..., 999.

255


Add REPEAT SxR Displays a dialog where a different, non-sequential, type of repeat can be defined. A new entry in the list is created.

REPEAT SxR takes a source id variable with a list of source-ids, a receiver id variable with a list or receiver-ids and loops through all combinations of source and receiver ids given. This is especially useful in cases where both many sources and many receivers are used, or when source or receiver numbering is not sequential. It is us used the same way as REPEAT so that if the sub-sequent step is Post-processing or Convolution the source and receiver variables are expanded to create the file names desired, such as E__.ECH. The list of source and receiver ids can be comma or space separated and for source ids constructs like: A* (meaning A0, A1, ..., A9) can be used and for receiver ids: 2* (meaning 20, 21, ..., 29). To include all receivers 0..9 use 0*. The following REPEAT SxR REPEAT-SxR SS A0 A1 RR 1 23 Post-processing C:\TEMP\E__.ECH C:\BIN.PST

will perform the same as: Post-processing Post-processing Post-processing Post-processing

C:\TEMP\E_A0_01.ECH C:\TEMP\E_A0_23.ECH C:\TEMP\E_A1_01.ECH C:\TEMP\E_A1_23.ECH

C:\BIN.PST C:\BIN.PST C:\BIN.PST C:\BIN.PST

256


Prediction... (Add processing steps) Displays a dialog allowing for a prediction PRD-file to be selected. A new entry in the list is created.

The Exec command item offers an opportunity to run a program or a script after each prediction, the sequence does not continue until the program/script is finished: (NONE)

(or an empty field) no program or script is run.

exec_matlab("[path]","script")

executes a [MATLAB] script, if MATLAB is not running it is started automatically. The first argument can give an optional path and an addpath command is sent before the MATLAB script is called so that an m-file given in script can be found (if not given, only m-files in the MATLAB startup path will be found). Note: a new MATLAB session is started so any variables have to be saved to be available next time. Examples: exec_matlab("","a = optimizeRT") exec_matlab("c:\mfiles","a = optimizeRT")

The results of the MATLAB script (if any) will be displayed in the *DEBUG* window. Note: the purpose of this exec_matlab is to calculate only, see more below, and no MATLAB graphics can be displayed. Example of a script (that does nothing useful) assumed to be c:\mfiles\testexec.m: function [a] = testexec(b) a = b + 2; return

This script would be called like: exec_matlab("c:\mfiles","a = testexec(2)")

and the result is shown in the *DEBUG* window: "exec_matlab_result: 4"

257

8. SEQUENCE PROCESSING MODULE exec_winexe("exe","options")

executes a Win32 application (.EXE), the full path must be included in exe. Any command-line options the executable offers can be given in options. Examples: exec_winexe("c:\progs\optimize.exe","-tolerance:2") exec_winexe("c:\catt_v8\cattplt32.exe","c:\out\RT_A0.PLT")

were the latter simply will display a result PLT-file in the standalone PLT-viewer and when the viewer is closed the sequence continues. exec_winexe("COMMAND","cmd")

executes cmd as if entered at a command prompt. Examples: exec_winexe("COMMAND","dir c:\temp\*.PLT /p") exec_winexe("COMMAND","c:\progs\renameplts.bat") exec_rexx("[path]","script_with_options")

executes a Rexx script (http://www.oorexx.org) assuming Rexx is installed. It performs the same action as if: > rexx script_with_options had been entered at a command prompt. If path is given it will be made current folder before the Rexx script is run. An example of a Rexx script (that does nothing useful) assumed to be c:\scripts\testexec.rex: say 'Please enter your name.' /* Display a message */ pull name /* Read response */ say 'Hello' name /* Display greeting */ exit 0 /* Exit with a return code of 0 */

This script would be called like: exec_rexx("c:\scripts","testexec.rex")

Further script languages may be added in future versions. General exec interface notes and tips

The basic idea with the exec interface is that a Prediction step can be run in a REPEAT loop so that after each prediction the given program or script is run and reads the latest results, rewrites a part of the geometry-, source- or receiver-files based on some criteria, and then a new calculation is started. A function can optionally return a result indicating if the Sequence can be stopped or not. At each step a function can also save the settings and selected results to give a history of the optimization. Examples of changes are: absorption and scattering coefficients, reflector angles, source aim, etc. Most practical is to gather the variable settings in a separate small GEO-file that acts as Master-file and that INCLUDEs the remaining geometry. The results for each run can be read from the output text-files such as PARAM_ss.TXT or PARAM_ssX.TXT but there is a risk that the format of these files will change between versions. Better is to, via a hidden option, select that Full detailed calculation will save all its prediction results also in binary MATLAB v4 format file (PARAM_ss.MAT). This file can be read from MATLAB with one command. The v4 format is simple to read also from an EXE program,

258

8. SEQUENCE PROCESSING MODULE contact CATT for details on the format. Since diffuse reflection is handled with a stochastic process no two runs will give identical results even if no input data is changed. This means that criteria must not be set too tight but with reasonable margins. A typical example would be to aim for a T-30 that is in a range say 1.5-1.7 rather than to aim for a T-30 = 1.6. The acoustic properties of a room has many degrees of freedom which means that optimization can be very difficult. Cases with the best prognosis are those where just a few parameters are changed. Variables can be used in the strings passed to the exec_ functions and will be expanded before calling. It is not necessary to always use an advanced optimization algorithm, the exec interface can also be used simply to run through a set of values for one or more settings and afterwards inspect the results and choose one of the combinations. Inside a function or script main results can either be stored in a file that is updated each run or the script can issue some system calls to rename results files or complete folders so that all results can be inspected afterwards. The exec_winexe example above calling the PLT-viewer can be used to manually alter some values. After the results have been viewed CATT-Edit can be used to change e.g. an absorption coefficient for another run. Post-processing... (Add processing steps) Displays a dialog allowing for a post-processing PST-file and an echogram ECH-file to be selected. A new entry in the list is created. Note: When post-proc. using a PST-file for CATTWalker CWI-generation no ECH-file needs to be selected, enter (NONE).

Mono… (Convolution) (Add processing steps) Displays a dialog allowing for an anechoic mono WAV-file, left and right binaural impulse response SIM-files (created by the Post-processing module) and a desired WAV-file to be created. If (NONE) is entered instead of the right SIM-file, mono convolution is performed of the left SIM-file with the anechoic WAV-file. A new entry in the list is created.

259


Left/Right/R+L Mix... (Convolution) (Add processing steps) Similar to the one above but instead a stereo anechoic file is entered and either the left, the right or a left+right mix is used as the input to the convolver. WAV-calibration... (Add processing steps) Displays a dialog allowing for a WAV-file to be selected as a template for calibration. A new entry in the list is created.

260


X-talk cancel... (Add processing steps) Displays a dialog allowing crosstalk cancellation of responses based on the latest saved settings (in Post-processing:Utilities|Crosstalk Canceling). A new entry in the list is created.

Headphone eq... (Add processing steps) Displays a dialog allowing for a headphone filter and a left or right SIM-file to be selected (both left and right files are filtered). A new entry in the list is created.

This is a streamlined version of the more general SIM-file filtering utility. The created files will automatically be renamed by changing the first letter in the name to "P" (Phones) unless the letter is "H". If the names start with "H" it is likely that the files have come from a HeadScape file generation and HeadScape can only read files with a fixed syntax starting with H. If the original, non-filtered, files are to be kept they must be duplicated in another folder before processing. SIM-file addition... (Add processing steps) Displays a dialog allowing for SIM-file addition of a large number of SIM-files in one step. Unlike when using the Utilities dialog for addition, the sequence method does not allow for change of gain or delay but uses the original ones.

261


The Template SIM-file selected determines that receiver used (01 in the example above), the folder and the addition compatibility. SIM-files based on the Source ids given will be used for addition. Source ids can be space or comma separated and B* means all B sources. In the example above the following SIM-files in C:\TEMP\OUT would be added: E_A0_01, E_A1_01 and E_B0_01, E_B1_01, E_B2_01, E_B3_01, E_B4_01, E_B5_01, E_B6_01, E_B7_01, E_B8_01, E_B9_01. Delete / Edit … Deletes / Edits the selected entry. Clone... Clones the currently selected entry and loads it for editing. If OK is pressed after editing, the new entry is inserted after the current one. This is a handy way to add e.g. several sequential post-processing steps where only the ECH-file differs with a receiver id, or several sequential convolutions where SIM-file and output WAV-file names differ only slightly. Move up / Move down Moves the selected entry one step up or down in the sequence.

262

9. CATT EDITOR

9. CATT EDITOR The CATT editor CATT-Edit (CATTED32.EXE) is essentially an enhanced Notepad clone that also offers search/replace, font selection, a Templates menu to insert various geometry-file templates, a GEO-file syntax reference, and an interface to the main help-file. Being a Notepad clone the editor suffers from similar limitations in file sizes it can handle under Windows 95/98/ME. The maximum file size is then typically 64 kB but for NT/2000/XP the limit is set to 1 MB. 64 kB may seem like a severe limitation, but the way the geometry-file format is organized it is bad practice to put all the data in one file. Better to divide it in logical parts and use the INCLUDE directive. It is only when a large file has been created by the AutoCAD interface that the geometry is described in a large single block and another editor must be used. But even then it is quite easy to divide the file into parts (e.g. corners in the master-file and planes in an included file) so that CATT-Edit can be used. The geometry check debug-file DEBUG.TXT may often be > 64 kB but may be made smaller by omitting some information such as the long list of plane data and/or corners only leaving the error checks. Instances of CATT-Edit and the main program communicates so that files can only be opened and edited in one instance of the editor. If a file that already is edited is selected for editing, the editor containing the file is moved to the front instead of loading a new instance of the editor. The communication works both when an editor is loaded from the main program and if it is loaded from e.g. Explorer. Similarly, if a file that is needed for a prediction (e.g. a GEO-file) is loaded in an instance of the editor without having been saved, the editor is moved to the front of other windows and a prompt is presented with a Yes/No/Cancel choice. Yes saves the file before proceeding, No proceeds without saving (i.e. uses the previous version on disk), and Cancel aborts the processing. Tab stops are initially set at every 5 average character widths of the selected font.

9.1 Non-standard menus File|Convert to ANSI

The editor allows for character set conversion so that files created by a DOS editor using the MS-DOS character set can be converted to the Windows ANSI set. Edit|Goto Line Number

For some geometry-file errors, the prediction module gives the error line number. This menu can be used to go directly to that line. The current line is displayed in the status-bar. Character|Font

By default the editor uses the font and font size that is selected in the main program File|Preferences. Character|Reduce/Enlarge Font

Reduces or enlarges the font used.

263

9. CATT EDITOR

Character|Increase/Decrease TAB-width

Increases or decreases the TAB-width by one average character. Template|

Place the cursor where e.g. a plane definition template (Plane [..) is wanted and select the corresponding menu item. The template will be created and the cursor is placed at a logical starting-position. Only functions with complex and/or hard-to-remember syntax have templates. For other commands see Help|Syntax below. Template|ABS Exchange

Inserts a record template for the CATT absorption library exchange file format. Useful if some data is entered directly in a text-file that can be imported, see Section 3.1 Template|RGB Color

Inserts, after visual color selection, an RGB triple (e.g. {255 0 0} for pure red) to be used in GEO-files. Template|loop(), recloop(), recloop2(), recwalk()

Inserts loop and similar templates. Template|Markers>

Inserts various marker templates. Template|Source>

Inserts various source syntax templates. Help|Help...

Loads the main help-file and displays the Prediction input-files topic. Help|GEO/LOC Syntax...

Displays a dialog with short syntax descriptions for GEO-, source- and receiver-file directives and tools. This dialog can stay open while editing. Help|Marker Syntax...

Displays a dialog with short syntax descriptions for Markers (visual non-acoustical elements in the optional GEO-file MARKERS section see Section 2.3). This dialog can stay open while editing.

264

10. AUTOCAD INTERFACE

10. AutoCAD INTERFACE 10.1 Introduction The direct AutoCAD interface for CATT-Acoustic consists of a set of AutoLISP procedures that create commands to be used within AutoCAD. These procedures then translates and writes a specific layer of an AutoCAD drawing to the CATT format. The manual describes v3.2 of the interface. To enable conversion in the opposite direction, the Prediction module can create an intermediate file that can be imported into AutoCAD by the interface. The conversion is carried out at the Prediction module menu selection File|Export Geometry To|AutoCAD interface CAD see Section 2.1. The interface allows for the complete geometry to be modeled inside AutoCAD including all surface planes (including correct normal definition), sources (position, aim, directivity, and Lp1m), receiver positions, plane names, and absorption names. The interface then writes a geometry-file, a source-, and a receiver-file with selected names. After that process, only some minor editing of absorption data in the GEO-file remains. It is, however, strongly recommended to create a CATT debug list-file and a PLINFO.PLT plot-file to check the model. The interface requires the full 3D version of AutoCAD (release 11 or later). This documentation naturally assumes that users have a basic understanding of the operation of AutoCAD. The interface was originally written by Advanced Engineering Computations (AEC), Gothenburg, Sweden and then gradually enhanced. As an alternative to using the AutoLISP interface the external CATT tool DXF2GEO (see File|External CATT Tool menu) can be used to convert from DXF to GEO. DXF2GEO comes with its own help-file but many of the discussions here about making models in rd AutoCAD apply for DXF as well. In addition 3 party tools to use other CAD programs such as SkecthUp may be available, see the users’ page (see Help|About for a direct link).

Direct creation of a model in AutoCAD For users that already know AutoCAD well, this is a straightforward approach potentially speeding up the modeling process considerably. This approach will not have any adverse effects when it comes to how appropriate the model will be for acoustic prediction using geometrical acoustics. The main difference between direct modeling in the CATT GEO-file format is that the model typically will consist of more planes thus making it appear more cluttered and make predictions somewhat slower. The reason for this is that the AutoCAD "3DFACE" element can have 3 or 4 vertices only while the GEO-file format allows for an unlimited number of plane vertices (corners). If the geometry is OK and no further re-import into AutoCAD is planned some surfaces, if desired, planes can be merged after import using File|Export Geometry To|Merged Planes GEO-file. If no alterations of the GEO-file created are made, a re-import of the files via the intermediate CAD-format should result in a model identical to the one that was exported.

265


Modification of an existing drawing in AutoCAD format To use an already existing AutoCAD drawing, e.g. provided by an architect, is potentially the most time-saving approach to modeling. It is also, however, associated with the greatest risks when it comes to how appropriate the model may be for room acoustic prediction based on geometrical acoustics. In most cases the original model needs to be simplified so that too small surfaces are removed and replaced by a frequency dependent scattering coefficient. For example, if audience chairs are individually modeled, they should be removed and replaced by a walled-in audience surface with appropriate absorption and scattering coefficients.

Import of an existing model in the CATT format In addition to re-importing models originally created in AutoCAD as discussed above, it may be interesting to import an existing model in the GEO-file format to render the model. Another example is when the structured GEO-file format has been selected because its features and possibilities for parametrical modeling using equations etc. If a model is created in GEO-file format directly without any thoughts of how AutoCAD handles surfaces, some features cannot be completely transported to AutoCAD. Examples of such features are planes that are subdivided into parts with different absorption, concave planes with many corners (due to an apparent bug in at least AutoCAD release 12 for Windows, these may not be handled well). Planes with hard-to-translate features are marked so that they can be checked and taken care of manually inside AutoCAD, see below. To create a file that can be imported in AutoCAD, the Prediction module menu item File|Export to CAD is selected. The current settings are inspected but only a few of them are actually used for the export: Output and Input folders, the Master-file, the Receiver-file, the Source-file, sources and receivers used, and Plane use settings. The created CAD-file contains geometry, source-, and receiver data in an easy-to-parse text format. Even if features in the GEO-file format, such as e.g. equations, have been used, the CAD-file contains only numbers, the unit is always meter and decimal point is always used. The format of a CAD-file is: comments ... (more comments) %SURFACEPROPERTIES …optional section when exporting for other uses than with the AutoCAD interface, see section 2.1 and File|Export Geometry To|AutoCAD Interface CAD ... %CORNERS co_id co_x co_y co_z ... (more corners) %PLANES pl_id / pl_name / pl_absname co_id1 co_id2 ... co_idN ... (more planes) %SOURCES s_id s_dirname:dirtype delay (e.g. A0 HORN12:SD1 20) s_x s_y s_z s_aim_x s_aim_y s_aim_z s_roll L125 L250 L500 L1k L2k L4k : L8k L16k ... (more sources) %RECEIVERS r_id r_x r_y r_z ... (more receivers) %EOF

266

10. AUTOCAD INTERFACE Example: CATT-Acoustic v8.0a, Copyright © CATT 1988-2002 CATT project to intermediate AutoCAD-format conversion (for >= v7.0) PROJECT=v8 Shoebox Note 1: plane sub-divisions are not written. Instead the full plane is written with abs #DIVIDED. Sub-divisions currently have to be handled as separate planes in AutoCAD. Note 2: Concave planes (with reentrant angles) are problematic to handle inside AutoCAD. Such planes are written with a # inserted as the first character in the absorption name. Note 3: abs/diff properties that are entered directly in a plane (Example: [1 floor / 1 2 3 4 / <8 7 6 5 4 3> ]) are assigned a dummy name starting with ## and then a number. Use names! Note 4: From 1, 2 and 3: all abs names starting with # need to be fixed! They can be transferred to a layer CATT_FIX with the command MARKFIX so that they can be taken care of. This will seldom be required for models originally created in AutoCAD. Note 5: All coordinate values are in meter. Note 6: The directivity name is set as name:type (e.g. SINGER:SD0). Note 7: The optional delay is always written (0.0 if not used). Note 8: The optional source roll is always written (0.0 if not used). Note 9: The optional 8k and 16k source Lp1m values are always written. %CORNERS 1 2 3 4 11 12 13 14

-5.000 -5.000 5.000 5.000 -5.000 -5.000 5.000 5.000

0.000 24.000 24.000 0.000 0.000 24.000 24.000 0.000

0.000 0.000 0.000 0.000 8.000 8.000 8.000 8.000

%PLANES 1 /floor/ AUDIENCE 4 3 2 1 2 /ceiling/ WOOD 11 12 13 14 3 /stage wall/ WOOD 1 11 14 4 4 /rear wall/ WOOD 3 13 12 2 5 /left wall/ WOOD 2 12 11 1 6 /right wall/ WOOD 4 14 13 3 %SOURCES A0 OMNI:SD0 1.000 1.000 70.0 73.0

0.00 1.700 3.700 76.0 79.0

1.700 1.700 82.0

95.0

0.000 : 95.0

95.0

%RECEIVERS 1 2

-3.000 3.000

20.000 20.000

1.300 1.300

%EOF

Absorption names for planes that could not be translated, and therefore have to be checked, are marked by a pre-pended # character. #DIVIDED means that the plane was divided in sub-planes and need to be manually fixed in AutoCAD. ##NNNN where NNNN is a number means that the absorption was entered with numbers directly instead of an absorption name. #ABSNAME where ABSNAME is an absorption name means that the plane has been concave

267

10. AUTOCAD INTERFACE and needs to be checked inside AutoCAD (the AutoCAD bug mentioned above may have created the set of 3DFACEs to represent the plane in a manner that is not correct). All planes that have absorption names prefixed with # can be marked by the provided AutoCAD command MARKFIX so that they can be taken care of.

Basic interface concepts The AutoCAD interface handles 3D elements of the types 3DFACE and 3DMESH. 3DMESH surfaces must be converted to 3DFACEs by the provided command CONV and plane and absorption names must be set. If desired, planes can be merged after import using the Prediction File|Export Geometry To|Merged Planes GEO-file. The interface works with three layers: CATT_MAIN, CATT_FIX, and CATT_TMP. When importing or exporting data the layer CATT_MAIN is used. CATT_FIX is used for planes that need to be "fixed" after import i.e. those whose absorption names start with #. CATT_TMP is for internal use. Generally, the interface does not perform extensive syntax check on e.g. valid source directivity or absorption names. This check is instead performed when the exported files are read by the Prediction module. Such errors are typically easily spotted and can be corrected with the editor. AutoCAD allows for the Space bar to act as Enter. The interface supports this except for when plane names and Lp1m values for the source are entered since they typically contain spaces.

10.2 Installation The procedure below describes v3.2 of the interface. The interface AutoLISP files (*.LSP) and predefined symbols (*.DWG) must be placed in a single folder. At installation they are placed in the CATT-folder ACAD. Start AutoCAD and load the AutoLISP command: (LOAD: "CATT-folder/ACAD/CA_LOAD")

Note: AutoCAD uses "/" or "\\" internally in path names. A perhaps more convenient alternative to the AutoLISP (LOAD ... ) command to load the interface command-set is to use the APPLOAD command giving a dialog to select the LSP-file to load. When the interface is run it will also ask for the location of the interface files. For frequent users of the interface it can be convenient to put this line in the acad.lsp file run at startup and no folder have to be selected: (SETENV "CaPath" ACAD-folder)

where ACAD-folder is the location of the interface files including ending / or \\.

10.3 AutoLISP commands Below is a list of the AutoLISP commands provided. Generally these commands work as any

268

10. AUTOCAD INTERFACE other AutoCAD commands in the sense that, for example, coordinates may be entered either by clicking with the mouse or by entering numbers at the command prompt. They can also act on a group of faces selected by the mouse.

SETCATTSCALE Sets or changes the unit and the scale for the drawing to be converted. This command must be used before any of the commands EXP, IMP, SRC, REC, CHDIRECT, or SHDIRECT can be used. If not, an error message will notify the user. With decimal units the scale is set so that numbers in the created GEO-file have the same magnitudes as those used inside AutoCAD. This is accomplished by the GEO-file directive SCALE. CATTSCALE

Displays current CATT unit and scale.

DRAW

Draws 3DFACEs in normal AutoCAD manner. This is just a replicate of the normal AutoCAD 3DFACE drawing command.

PUT

Puts plane and absorption names on currently selected 3DFACEs and moves them from the current layer to the CATT_MAIN layer.

GET

Gets and shows plane and absorption name on currently selected 3DFACEs.

CONV

Converts a 3DMESH to 3DFACEs (i.e. explodes the mesh). Sets plane and absorption names on the created 3DFACEs and moves them from the current layer to the CATT_MAIN layer.

SRC

Inserts a source in the drawing. Asks for source id, position coordinate, aim coordinate, source name (i.e. directivity), and Lp1m_a values. The source type is assumed electro-acoustic and Gain_a is set to < 0 … >. CATTSCALE must be set before SRC is used.

CHSRCPOS

Changes the source position. For an old DRW-file made using v2 of the interface, the source has to be removed and entered again to make this command work (an old source will export well but the position cannot be changed with this command).

CHSRCAIM

Changes the source aim point. For an old DRW-file made using v2 of the interface, the source has to be removed and entered again to make this command work (an old source will export well but the aim cannot be changed with this command).

REC

Inserts a receiver in the drawing. Asks for receiver id and position coordinate. CATTSCALE must be set before REC is used.

EXP

Exports all 3DFACEs, sources, and receivers on the layer CATT_MAIN to selected source, receiver, and GEO-files. If the model contains many surfaces the process may take a couple of minutes. If no sources or receivers have been positioned (using SRC and REC), the corresponding files will not be created. If name and absorption names have not been set on some faces, default values are set so that, when the files are read by the prediction module, they have a valid format. The interface declares all absorption names used in the model in the directives section of the created GEO-file. Some minor editing is then required to add actual numeric values for absorption and diffusion factors. If, however, the material is in the Surface properties library the corresponding declaration can simply be deleted or commented out. If all absorption names have not been given inside AutoCAD, it has to be corrected for by substituting the default absorption for a real one. CATTSCALE must be set before EXP is used. If desired, planes can be merged after import using the Prediction

269

10. AUTOCAD INTERFACE File|Export Geometry To|Merged Planes GEO-file.

IMP

Imports to the layer CATT_MAIN geometry from the intermediate CAD-file format created by the prediction File|Export to CAD menu selection. 3DFACEs are created for all planes thereby splitting surfaces with more than 4 corners. All imported planes that need manual attention have an absorption name starting with #. These faces can be moved to the layer CATT_FIX by the command MARKFIX. CATTSCALE must be set before IMP is used.

MARKFIX

Marks 3DFACEs that need to be fixed. MARKFIX should be used after IMP. All 3DFACEs with an absorption name starting with # are moved to the layer CATT_FIX. If, after export, some absorption names still start with #, the prediction module will issue a syntax error pointing out the problem plane definition.

UNMARKFIX

Unmarks selected and fixed 3DFACEs and moves them from CATT_FIX to CATT_MAIN. To be used when faces have been corrected after import.

SHDIV

Shows division of a surface. After import planes with many corners have been subdivided into planes with 3-4 corners. This command shows the division lines. In many cases a more logical, and visually more pleasing, division may be created manually. SHDIV is actually just a toggle on/off of the AutoCAD system variable SPLFRAME.

HIDEDIV

Hide division of a surface.

SHDIRECT

Shows direction of plane normals on selected faces. The normal direction is shown indirectly by the absorption name in 3D text on a hashed surface. If the text can be read normally the face is viewed from its reflecting side, if the text is mirrored the face is viewed from its back side. The text and the hash pattern is located on the CATT_TMP layer and is never exported. CATTSCALE must be set before SHDIRECT is used.

SHR

A simpler version of SHDIRECT only showing the letter "R" on selected surfaces. If R can be read correctly (i.e. not mirrored) the plane is viewed from its Reflecting side. The R symbols are removed with SHDIRECT. After a surface is DRAWn, a SHR command is automatically issued made for the newly created surfaces.

DELSHDIRECT

Removes "Show direction" of plane normals on selected faces.

CHDIRECT

Change direction of plane normals on selected faces. Also performs DELSHDIRECT on not selected surfaces. CATTSCALE must be set before CHDIRECT is used.

The AutoCAD command DDATTE can be used to display source data (type DDATTE and Enter and then click on the source). The AutoCAD commands MOVE and STRETCH can be used to manipulate a source pos-aim group (requires that the source is created in v3 of the interface).

270


10.4 Modeling tips If only half of a symmetrical model is created in AutoCAD the MIRROR GEO-file directive can be added to the created GEO-file and maybe also a smaller number of bridging planes may be needed to be created "manually" but all corners would already be defined. When selecting planes it is often difficult to select the edge of the plane one wants (the neighbour plane gets selected instead). For this a standard AutoCAD feature can be used: Press and hold the Ctrl key while Left button clicking on the edge and the objects with the common edge will alternate.

271


272

11. TUTORIAL

11. TUTORIAL This section describes one way to model a hall using CATT-Acoustic v8.0. Since it is a tutorial, most tools provided by the software are exemplified (except Markers since they are easy to use once the GEO-file concept is understood). The GEO-format is very flexible and several different procedures can be followed when modeling a hall. Even when preferring to create the geometry in a CAD program (see section 10) there are many useful features in the GEOformat that can be applied after export to GEO. All steps from modeling to prediction, source addition, post-processing and convolution are exemplified. The tutorial was installed in the CATTDATA-folder MODELS\TUTORIAL. It contains three ready-made geometry-files, a source-file, a receiver-file, and a basic Prediction settings-file: MASTSTEP.GEO

contains the complete geometry and each tutorial step can be selected by setting a constant STEP, at a screen prompt, to the value 1 to 6 corresponding to the tutorial steps.

MASTFULL.GEO

contains the complete geometry where each tutorial step is commented.

REFLECT.GEO

defines the reflectors used in step 6.

SRC.LOC

is the source-file.

REC.LOC

is the receiver-file.

GEO.PRD

is a Prediction settings-file.

MASTFULL.GEO, SRC.LOC, and REC.LOC can also be seen in Appendix A.

Note that the help-file version of the tutorial actually includes most essential plot-files created so that the whole tutorial can be initially "run" inside the help-file.

11.1 Preparation Before one starts entering the actual data for a new hall, it is useful to draw a simple 3D outline of the hall, at last for the initial planes to be created. As the model progresses one can instead use plots generated by the software. If the hall to be modeled already exists and is going to be renovated it is extremely valuable with photos from the hall since drawings are not always accurate, up to date and/or may be difficult to interpret. Photos are also of great help when deciding on surface properties.

11.2 Modeling The steps involved in a new project are: Create a new project, create the geometry, and finally perform the prediction. If desired, then also Source addition and/or Post-processing for auralization can be performed. Start CATT-Acoustic. The Prediction module is initially open and active. The View module is also open but no plot-file is loaded. When creating a new project from scratch (but not here in the tutorial) the following steps must be performed. First select Utilities|Create a New Project.

273

11. TUTORIAL

Enter a project folder and a project name for reference. If the specified folder does not exist it will be created. The Browse button also allows for creating a folder. For a new project, files such as those below are created in the project folder when OK is pressed.

MASTER.GEO (template master-file for the geometry): ;MASTER.GEO ;PROJECT:Sample project ;INCLUDE ;ABS absname = <10 10 10 10 10 10> ;L <10 10 10 10 10 10> CORNERS ;id x y z PLANES ;[id name / / absname ] ;[id name / / (a / / a_abs) (b / / b_abs) ]

SRC.LOC (template source location and data file): ;PROJECT= Sample project SOURCEDEFS ; a natural source ;id position directivity aim-position [roll] A0 0.0 3.0 1.7 OMNI 0.0 5.0 1.7 Lp1m_a = Lp_white 94 ; white spectrum, 94 dB at 1kHz ; an electro-acoustical source (syntax 1 specifying Gain_a) B0 -2.0 3.0 3.7 Catt.SD0 0.0 7.0 1.7 Lp1m_a = Lp_voice_normal Gain_a = <20 20 20 20 20 20> ; calculates Lp1m_ea Delay_e = 0 ; an electro-acoustical source (syntax 2 specifying Lp1m_ea) B1 2.0 3.0 3.7 Catt.SD0 0.0 7.0 1.7 Lp1m_a = Lp_voice_normal Lp1m_ea = <75 80 82 85 88 88> ; at 1m on source axis Delay_e = 0

REC.LOC (template receiver location file): ;REC.LOC ;PROJECT: Sample project RECEIVERS 1 0.0 10.0 3.0

PROJECT.TXT (optionally to be used for project log): Project: Sample project Sign. : Date : Log :

A basic prediction settings-file GEO.PRD is also created prepared for geometry check. For a new project these data-files must then be filled in using the editor, exported via the AutoCAD interface, via DXF2GEO or a modeling plugin. In this tutorial however, a CATTDATA-folder MODELS\TUTORIAL is already created and all project files already exist. In the rest of the tutorial the geometry data does not have to be entered. Just read the paragraphs, make changes to the settings-files as described, run the Prediction, enter the appropriate STEP value when prompted and view the various plot- and result-files created.

274

11. TUTORIAL Select File|Open Settings and select CATTDATA-folder MODELS\TUTORIAL\GEO.PRD. The PRD-file uses a shortcut for Input and Output folder in the General settings dialog where "." means the same folder where the PRD resides, in other cases it may be chosen to enter absolute paths here. Check that Master-file is set to MASTSTEP.GEO. This master GEO-file contains the complete tutorial geometry and later when Save and Run is pressed, STEP = 1, 2, 3, 4, 5 or 6 can be selected at the pop-up dialog corresponding to the steps described in this tutorial. Browse through the rest of the settings but leave them as they are for now. Press OK and leave General settings and then select Geometry view/check and toggle all Plot-files on except SRCINFO and REFLINFO. Select (Primary 3D view) Natural and View distance to 65 m (the 3D view can be changed interactively when the plot-file is shown, this defines only the initial setting). Press OK and leave Geometry view/check. Below are shown six steps towards a complete model. For each step only the additional GEO data for the step is shown. For the final complete GEO-, source-, and receiver-files refer to Appendix A. Remember - you do not need to enter the geometry data yourself. Just read the paragraphs, press Save and Run, enter STEP and view the various plot-files (some are shown here and some more are included in the help-file).

Step 1, the audience and floor surfaces First to be entered is the floor and audience coordinates planes are entered (the complete GEO-file can be viewed by selecting (Edit) Geo-file on the main dialog and opening MASTSTEP.GEO or MASTFULL.GEO or by checking Appendix A). The tutorial input-files uses decimal point but both decimal comma and point are allowed. A "walled in" audience is created where audience or seat absorption is applied also to the sides, see Fig. 2.8. "Help corners" are created to enable precise locking of receiver locations at a fixed height above the audience (step 5). Note that the audience has been assigned specific scattering coefficients. For this simple project remaining surfaces all get default scattering coefficients introduced later. ;MASTFULL.GEO ;PROJECT:CATT-Acoustic tutorial for v8.0 ;general materials defined here or from library ABS wood = <12 10 8 7 5 4> {213 220 160} ABS carpet = < 10 8 6 4 3 2 > {98 209 44} ABS audience = <35 50 55 70 85 80> L <30 40 50 60 70 80> {128 128 0} ABS floorabs = carpet GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL

ah rh aww fsd fsw sw pw w ew fld flt wl1

= = = = = = = = = = = =

1.0 0.3 1.0 1 6 fsw 2 16 1.4 3 1 6

; ; ; ; ; + ; ; ; ; ; ;

audience height receiver height above audience plane audience sidewalk widths front-stage depth front-stage width 2 ; stage width prosc. width max hall width entrance width flank depth floor tilt first side wall length

275

11. TUTORIAL GLOBAL wl2 = 7 ; second side wall length GLOBAL hd = fld + wl1 + wl2 ; hall depth from prosc MIRROR 100 200

; corner and plane offsets for mirrored part

CORNERS ;floor corners 1 -fsw/2 fsd 0 2 -sw/2 0 0 3 x(2)-pw 0 0 4 -w/2 fld 0 7 -w/2+ew hd flt 8 0 y(4) 0 ; help corner for lock 5 x(4) fld+wl1 lock(4 7 8) 6 x(7) y(5) lock(4 7 8) ;audience surface ;20-21 help corners to lock audience corners at ah above floor 20 x(8) y(8) z(8)+ah 21 x(4) y(4) z(4)+ah 22 x(7) y(7) z(7)+ah ;23-25 help corners to lock receiver positions at rh above audience ;surface (in REC.LOC) 23 x(20) y(20) z(20)+rh 24 x(21) y(21) z(21)+rh 25 x(22) y(22) z(22)+rh ;audience upper corners 30 x(4)+aww y(4) lock(20 31 x(4)+aww y(5)-aww lock(20 32 x(7)+aww y(5)-aww lock(20 33 x(7)+aww y(7)-aww lock(20 ;audience lower corners 40 x(30) y(30) lock(4 7 41 x(31) y(31) lock(4 7 42 x(32) y(32) lock(4 7 43 x(33) y(33) lock(4 7

21 21 21 21

22) 22) 22) 22)

8) 8) 8) 8)

PLANES ;audience surfaces [1 audience surface / 33 32 31 30 130 131 132 133 / audience ] [2 audience front / 140 130 30 40 / audience ] [3 audience rear / 143 43 33 133 / audience ] [4 audience front side / 30 31 41 40 / audience ] [5 audience middle side / 42 41 31 32 / audience ] [6 audience rear side / 43 42 32 33 / audience ] ;floor surfaces [7 front floor / 101 102 103 104 140 40 4 3 2 1 / floorabs ] [8 sloped side floor / 40 41 42 43 7 6 5 4 / floorabs ] [9 sloped rear floor / 107 7 43 143 / floorabs ]

The coordinate system must have the y-axis towards the audience and the z-axis upwards (gives the x-axis left to right as seen from the stage looking towards the audience) see Section 1.5. To utilize mirroring x = 0 should be placed at the symmetry line. The plane corners must be entered in clockwise order as seen from behind each surface (in fact, the direction is selectable but must be consistent in each file and flipping "/" to "\" reverses the direction). Note the MIRROR directive that creates mirror copies of all corners with 100 added to their ids (no planes are mirrored in this step since the floor plane coordinates contain both negative and positive x-values). The lock() function is used to lock hard-to-measure coordinates into a plane defined by three corners. Planes can then be made perfectly flat.

276

11. TUTORIAL

Note: if expressions are used for coordinate values, they must either be entered without blanks or enclosed in round brackets. An example from above: the y-value of corner 8 fld+wl1 could also have been entered as (fld + wl1). The first part of the geometry is now created. Source and receivers are of little concern at this stage. Press Save and Run. Enter 1 for STEP when the dialog appears. The geometry-file is read and requested plot-files are created in the output folder. The View module automatically shows all created plot-files (COLORED.PLT, SHADED.PLT, SHADED.OGL for the CATT 3D-viewer, COINFO.PLT see Fig. 11.1, PLINFO.PLT, VIEW4.PLT).

33 7 43 31 5 6 4132 42

30 4 40

01 133 143 107

A0

3 2

1

101 102 103

132 142131 106 141 105 130 140 104

Fig 11.1 COINFO.PLT after Step 1

The plot-files names are saved in PLT_GEO.TXT. On the Plot-file control dialog or toolbar select next/prev File to view the files ("<<" and ">>" buttons or press PgUp/PgDn). PLINFO.PLT contains plot "sub-frames" that are selected by the Frame buttons ("<" and ">"), the left-right arrow keys. To jump directly to a certain frame (plane), either choose Options|Select Frame or press "F" to toggle mouse selection on (press "F" again to toggle off and manipulate the view again). A plane should always have its reflecting side colored and the back black or the corner entering order must be reversed. This is very very important!

Step 2, walls and ceiling In this step all ceiling coordinates are entered and main walls and the ceiling are defined. Note: below are shown only the new lines associated with step 2. Again, for the full file refer to Appendix A. ABS ceilabs = wood ABS wallabs = wood GLOBAL h

= 8

; hall height

CORNERS ;ceiling 52 x(2) y(2) 53 x(3) y(3) 54 x(4) y(4) 55 x(5) y(5) 56 x(6) y(6)

h h h h h

277

11. TUTORIAL 57 x(7) y(7) h PLANES [20 [21 [22 [23 [24

ceiling / 52 53 54 55 56 57 157 156 155 154 153 152 / ceilabs ] back wall / 107 157 57 7 / wallabs ] side wall front / 5 55 54 4 / wallabs ] side wall back / 56 6 7 57 / wallabs ] front flank wall / 54 53 3 4 / wallabs ]

Note the use of x() and y() functions to copy x, and y-coordinates from the floor to the ceiling. If a floor coordinate is changed, the ceiling will automatically follow. The second part of the geometry is now created. Press Save and Run. Enter 2 for STEP when the dialog appears.

57 55

56

54 157 53 52

33 7 43 31 5 32 6 41 152 42

4

30 40

156 155

01 154

153

133 143 107

A0

3 2

132 106 142131 141 105

1

101 102

130 140 104

103

Fig 11.2 COINFO.PLT after Step 2

Step 3, entrance wall and door The entrance door is created as a plane sub-division where the door is the first division and the complete entrance wall is the second. To use plane sub-divisions instead of defining separate planes is more logical and more efficient. For details about plane sub-divisions see Section 2.3. ABS doorabs = wood GLOBAL e GLOBAL dh

= 0.2 ; edge width entrance door = 2 ; entrance door height

CORNERS ;entrance 60 x(5)+e 61 x(5)+e 62 x(6)-e 63 x(6)-e

door y(5) y(5) y(5) y(5)

z(5) z(5)+dh z(5)+dh z(5)

278

11. TUTORIAL PLANES [30 entrance wall / 5 6 56 55 / (door / 63 62 61 60 / doorabs) (wall / 5 6 56 55 / wallabs) ]

As seen from the above, plane-definitions can spread over several lines if required. Once again press Save and Run and enter for 3 for STEP in the dialog. door (

100%

30)

DOORABS <12 10

8

7

5

4 :

3

2> L <10 10 10 10 10 10 : 10 10> 50%

0%

01

61 62

60 63 A0

Fig 11.3 PLINFO.PLT after Step 3

Step 4, the stage The stage and the proscenium can now be connected. ABS sfloorab = wood ABS swallabs = wood GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL

sh sd sbw sfh sbh

= = = = =

1 ; 5 ; 4 ; h-2 ; sfh-1

stage height stage depth stage back width stage front height ; stage back height

CORNERS ;prosc. and stage 74 x(2) y(2) sh 75 x(1) y(1) sh 76 -sbw/2 -sd sh 77 x(76) y(76) sh+sbh 78 x(2) y(2) sh+sfh PLANES [40 [41 [42 [43 [44 [45

prosc. walls stage side edge stage front edge stage floor stage ceiling stage back wall

/ / / / / /

2 3 53 52 78 74 / wallabs ] 2 74 75 1 / swallabs ] 1 75 175 101 / swallabs ] 176 174 175 75 74 76 / sfloorab ] 77 78 178 177 / swallabs ] 76 77 177 176 / swallabs ]

279

125

500

2k

8k

11. TUTORIAL [46 stage side wall [47 prosc. top

/ 76 74 78 77 / swallabs ] / 78 52 152 178 / swallabs ]

Again press Save and Run and enter for 4 for STEP in the dialog.

01

A0

Fig 11.4 COLORED.PLT after Step 4

Step 5, source/receiver locations Source-file (SRC.LOC): "Omni" natural directivity selected. A second source is added since we will later add the results of the two sources together. ;SRC.LOC ;PROJECT:CATT-Acoustic tutorial v8.0 SOURCEDEFS IF step < 5 THEN A0 0.0 3.0 2.0 OMNI 0.0 10.0 3.0 Lp1m_a = <70 73 76 79 82 85> B0 1.0 2.0 2.0 OMNI 0.0 10.0 3.0 Lp1m_a = <70 73 76 79 82 85> ELSE A0 -2.5 fsd-0.5 sh+1.7 OMNI -2.0 fsd+2 sh+1.7 Lp1m_a = <70 73 76 79 82 85> B0 1.0 fsd-2.0 sh+1.7 OMNI Lp1m_a = <70 73 76 79 82 85> ENDIF

1.0 fsd+2 sh+1.7

280

11. TUTORIAL

Receiver-file (REC.LOC): ;REC.LOC ;PROJECT:CATT-Acoustic tutorial v8.0 IF step >= 5 THEN LOCAL dx = 2.2 LOCAL dy = 3 LOCAL y0 = fld+1 ENDIF RECEIVERS IF step < 5 THEN 1 0.0 10.0 3.0 ELSE 1 -2*dx y0 2 -1*dx y0 3 0*dx y0 4 1*dx y0 5 2*dx y0 11 -2*dx y0+dy 12 -1*dx y0+dy 13 0*dx y0+dy 14 1*dx y0+dy 15 2*dx y0+dy 21 -2*dx y0+2*dy 22 -1*dx y0+2*dy 23 0*dx y0+2*dy 24 1*dx y0+2*dy 25 2*dx y0+2*dy 31 -2*dx y0+3*dy 32 -1*dx y0+3*dy 33 0*dx y0+3*dy 34 1*dx y0+3*dy 35 2*dx y0+3*dy ENDIF

lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23

24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24

25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25)

Select General settings/Receivers and toggle 1-5, 11-15, 21-25, and 31-35 on. Select (Sources)Which to use and toggle both source A0 and B0 on. Set head-direction to Stage. Run again and enter 5 at the screen prompt.

31 32

21

A0 B0

25

14

03

15

04 05

Fig 11.5 SHADED.PLT after Step 5

281

35

24

13

02

34

23

12

01

33

22

11

11. TUTORIAL

Step 6, side wall reflectors Reflectors separate from the main geometry are best defined in separate files. Such files are declared as OBJECTs and can thereby be translated/rotated. The defined geometry is used as a template and COPY creates further identical copies with various translation/rotation values. In the top "directives" section in the master-file the reflector-file is included: ;STEP 6 - side wall reflectors INCLUDE reflect.geo

In a new GEO-file REFLECT.GEO: ;REFLECT.GEO ;PROJECT:CATT-Acoustic tutorial for v8.0 ; STEP 6 side-wall reflectors ABS backabs = <20 30 40 50 60 60> {64 128 128} ABS frontabs = <12 10 8 7 6 5> {0 0 128} LOCAL r_w LOCAL r_h

= wl1/5 = h-3

LOCAL tx = w/2 - 0.5 LOCAL ty = r_w LOCAL tz = 2.75 LOCAL ry = -10 LOCAL rz = 15 OBJECT ROTATE TRANSLATE OFFSETPL OFFSETCO

0 tx

ry fld+ty

rz tz

100 80

MIRROR 100 200 COPY 5 2 COPY 10 4 COPY 15 6

0 0 0 0 0 0 0 0 0

tx fld+2*ty tz tx fld+3*ty tz tx fld+4*ty tz

0 ry rz 0 ry rz 0 ry rz

CORNERS 0 1 2 3

0 -r_w/2 0 0 -r_w/2 r_h 0 r_w/2 r_h 0 r_w/2 0

PLANES ;* = auto edge [0 refl front ;the back MUST [1 refl back

diffusion + enable in General settings / 0 1 2 3 / frontabs* ] be defined too \ 0 1 2 3 \ backabs* ]

Note the use of OFFSETPL and OFFSETCO to make the OBJECT GEO-file somewhat independent. To re-use it in another project, only the offsets and the absorption need to be changed. Run again and enter 6 for STEP at the screen prompt, see Fig. 11.6 for the final design. For some idea of how the reflectors work, go to General settings and select the Refl. Info plot-file (REFLINFO.PLT) and specify the plane ids 100, 102, 104 and 106 (front sides of the right side reflectors). Run again and look at the reflector coverage.

282

11. TUTORIAL

Z

Z

Y

X

5m Audience:149m²

A0

01

11

21

31

02

12

22

32

03

13

23

33

04

14

24

34

05

15

25

35

B0

Volume:1769m³

Y

X

Fig 11.6 VIEW4.PLT after Step 6

11.3 Prediction The following steps may be taken in different order depending on the type of project. The Prediction module can create any combination of detailed early specular reflections, audience area mapping (direct sound and/or direct+reflected), full detailed calculation, and geometry check as well as direct sound Pixel rendering and Interactive RT estimation. For some projects the shape may be most natural to check first (mapping) but for others the reverberation time and details at specific receiver positions may be more natural to check first (full detailed calculation). First select General settings and select Master-file to MASTFULL.GEO so you do not need to select the STEP anymore. It is now also time to check/select (Diffuse reflection) on or off and if Diffuse defaults will be used. If Diffuse defaults is used then all materials without diffusion properties defined will get the selected default scattering coefficients. Diffuse reflections on/off can have a strong impact on the estimation of reverberation time, see Section 2.5. For this tutorial set default scattering coefficients to 30% for all bands. Note: Sabine and Eyring RT formulas assume 100% diffusion or a "mixing" room shape, don't be surprised if the estimated T-30 is higher, it is normal in these types of rooms. A good experiment is to set all scattering coefficients to 99% and see how T-30 become almost identical to the Eyring RT, another good experiment is to switch off diffuse reflection in General settings and see how T-30 becomes >> Eyring RT.

Interactive RT estimate A good start can be to estimate the classical reverberation time according to Sabine and Eyring RT and to perform a quick ray-trace for a global T-30 estimate. Click Interactive RT estimate and try changing some of the materials of each plane by clicking the property name button of the selected plane or sub-division. The Eyring and Sabine RT estimates are immediately updated, to run a quick global ray-trace click Start trace and watch a continuously updated T-30 curve.

283

11. TUTORIAL

Audience area mapping Leave General settings by pressing OK and select (Audience area mapping) Direct + reflected sound. Select No of rays/oct to 10000 (or Auto number for a more thorough calculation), Auto time, a Map step of 0.5 m and a Map height of 0.25. For Audience planes select no. 1 (to enter only one plane in the dialog enter 1 only in the left Range edit box or in both and click Add). Spherical receivers (slightly overlapping) will be automatically placed at the Map height over the audience plane. Audience planes will be marked grey in the VIEW4.PLT file. Browse through the rest of the settings since there are many mapping options (up to 12 2D or 3D parameter mapping plots plus RASTI, STI and early sound coverage). Leave Audience area mapping by pressing OK. Click Save and Run. The resulting plot-files will be for 1 kHz and for sound pressure level (SPL_1K.PLT) and lateral fraction (LF_1K.PLT) and parameters selected (PAR1_1K.PLT). Note that the results now are for source A0 and B0 added together.

D-50

[%] 1 kHz

A0 B0

LF

C-80 [dB] 1 kHz 60 55 50 45 40 35 30 25 20

[%] 1 kHz

A0 B0

4 3 A0 B0

2 1 0 -1 -2

Ts

[ms] 1 kHz

40

110

35

100

30

90

25 80

20 15

70

10

60

Fig. 11.7 PAR1_1K.PLT from Audience area mapping

Early part detailed ISM Select Early part detailed ISM on and Audience area mapping off. Edit Early part detailed ISM and set specular refl. orders to 3. Select Truncation time to 150 ms (the truncation time would typically range from 150 ms for small halls to 300 ms for large halls). Toggle Refl. history plot-file on, the rest off. Browse through the rest of the settings.

284

11. TUTORIAL dB 70

H_A0_05_1K

Receiver angles

45

60 50 40 30 20 10

0 23

150 ms 173

50 100 73 123 Source angles

Ver S-24-20-21-R

Hor

31 21 32 11 22 33 A0 01 12 23 34 02 13 24 35 B0 03 14 25 04 15 05 Ver

Hor

Fig. 11.8 H_A0_05_1K.PLT from Early part detailed ISM

Click Save and Run. The main part of the prediction time is spent saving detailed plot-files for reflection histories for each of the 20 receivers. The resulting plot-files will be for 1 kHz (H_A0_rr_1K.PLT, see Fig.11.8, and H_B0_rr_1K.PLT) where rr are the receiver ids.

Full detailed calculation Full detailed calculation, using the RTC, is the main prediction tool. Toggle Full detailed calculation on and Early part detailed ISM off. Edit Full detailed calculation and select No of rays/oct 1000 (or for a very thorough prediction to Auto). The Ray truncation time should be set to correspond roughly to the RT, say 1500 ms, or to Auto. Toggle on RT overview, Echograms, Sound roses, Vectorgrams, Decays, Parameter map and Parameter trend. Select both Save data for items for later use with Source addition and Post-processing (for auralization). Browse through the rest of the settings. Leave the dialog by pressing OK. Click Save and Run. Note that if e.g. 1000 rays were selected 8000 rays are actually traced (1000 per octave). The resulting plot-files will be: RT_A0.PLT E_A0_rr_INT E_A0_rr _1K R_A0_rr_1K V_A0_rr_1K PARAM1_A0.PLT TREND1_A0.PLT

RT_B0.PLT E_B0_rr_INT.PLT E_B0_rr_1K.PLT R_B0_rr_1K V_B0_rr_1K PARAM1_B0.PLT TREND1_B0.PLT

global reverberation time and other statistics decay curves all octaves detailed echograms and results, se Fig. 11.9 sound roses vectorgrams parameter overview parameter trend.

285

11. TUTORIAL dB 80

Complete echogram

dB 80

70

70 1.23s (-1.00) 1.25s (-1.00)

60 50

60 50

40

40

30

30

20

20

10

Early echogram

0

500

EDT T-15 T-30

1.14 s 1.23 s 1.25 s

D-50 C-80 LFC LF Ts

50.0 2.7 43.3 29.0 79.2

SPL G

71.4 dB 12.4 dB

1000

10

1500 ms

0 23 dB 80

4ms Filter

ms 50 100 150 200 73 123 173 223 Early directional echograms

70 60

% dB % % ms

Up

50 05

40

A0 OMNI DI = 0 dB 79.0 dB at 1 m 1kHz 5m

30

Lt

Rt Dn

20 10

0 23

50 73

100 123

150 173

200 223

ms

Fig. 11.9 E_A0_05_1K.PLT from Full detailed calculation

Since now also numerical results are interesting text-files are created. Select PARAM_A0.TXT or PARAM_B0.TXT from the Prediction:Latest Results list-box (double-click or select and press Enter) and it will load the file into CATT-Edit (in view-only mode). This text-file summarizes all results for room acoustic measures as well as how the prediction has been performed (no of rays etc.). Also files suitable for spreadsheet import (e.g. MS Excel) were created (PARAM_A0x.TXT and PARAM_B0x.TXT) where tables are semi-colon delimited and non-tabular data and text is quoted. The prediction is now complete. Go to Source addition (Section 11.4) or Post-processing (Section 11.5).

11.4 Source addition To add the detailed results of multiple sources together the (Save data for) Source addition option in the Full detailed calculation dialog (or for special cases Early part detailed ISM) must be set. The Prediction module then creates special echogram files with the name syntax E_ss_rr.ADD and a file called Name.CMB (where Name.LOC is the name of the current prediction source-file, typically SRC.CMB) containing source combination data (directivity type, optional delay, aim coordinate, Lp1m values, and a geometry checksum). Note: the Audience area mapping adds sources together automatically but mainly graphically presented results are created while the Source addition module enables individual point receiver echogram studies as well as source changes without a full recalculation. Open or select the Source addition module from the Windows menu. When the module opens select CATTDATA-folder MODELS\TUTORIAL\OUT\SRC.CMB that was created at prediction. When the CMB-file loads, all E_ss_rr.ADD files in the same folder as the CMBfile (typically a prediction output folder) are compared so that they all have the same geometry checksum. This is required since ADD-files may have been created at different times - one source at a time. All compatible sources found appear enabled in the (Sources) group. Similarly, if the Receivers button is pressed all compatible receivers appear enabled. Note that only the common set of source and receives are enabled (e.g. if A0 and 01, 02 have been predicted at one time, and B0 and 01, 02, 03 at another time, only A0, B0, 01, and 02 will be enabled).

286

11. TUTORIAL In the (Sources) group enabled sources can be switched on/off and edited by clicking on their respective check boxes and edit buttons. Keep the initial settings. Click the Receivers button and select only receiver 05 (note that also several receivers in the groups 10-19, 20-29 and 30-39 are enabled) and click OK. Click Save and run and the addition of the echograms for source A0 and B0 to receiver 5 will be calculated. This process only takes a couple of seconds and displays a plot-file E_01_05_1K.PLT, see Fig. 11-10, where both the individual A0->05 and B0->05 echograms (solid lines) and the total A0+B0->05 (dotted lines) are shown (the 01 number just tells that this is the first file in a possible sequence since when using 260 sources there will be 33 files numbered from E_01_rr_oct.PLT to E_33_rr_oct.PLT). The direct sound from each source is marked by a circle. A multiple source decay plot-file E_05_INT.PLT and a result text-file is also created with the most essential parameters, E_05.TXT. dB 70 60 50 40 30 20 dB 70 60 50 40 30 20

Fig. 11.10

A0 0.0ms OMNI.SD0

0

100

0

100

ms 200 B0 0.0ms OMNI.SD0

200

ms

E_01_05_1K.PLT from Source addition.

Now it is time for experimenting with the two sources. Click an edit button for source B and select another source directivity (CATT.SD0) and set a delay of 20 ms. Click Save and run again and watch how the B0->05 echogram is shifted by 20 ms and details have changed because of the new directivity is applied, see Fig. 11.11. dB 70 60 50 40 30 20 dB 70 60 50 40 30 20

A0 0.0ms OMNI.SD0

0

100

0

100

ms 200 B0 20.0ms CATT.SD0

200

ms

Fig. 11.11 E_01_05_1K.PLT from Source add., B0 delayed 20 ms and with a new directivity.

Note that these source changes did not require a full recalculation in the Prediction module. Click the Receivers button and select all the receivers, then click OK. Click Save and run and the addition of the echograms for source A0 and B0 to all receivers will be calculated. When also multiple receivers are calculated for, the results are a bit different and have the name syntax PARAM.TXT and TREND1.PLT and TREND2.PLT similar to the PARAM_ss.TXT and TREND_ss.PLT files created by the Prediction module.

287

11. TUTORIAL

11.5 Post-processing - Auralization To create an auralization of the results from a prediction, a (Save data for) Post-processing option in the Full detailed calculation dialog (or for special cases Early part detailed ISM) must be set. The Prediction module then creates special echogram files with the name syntax E_ss_rr.ECH. Optionally, ECH-files can be created (or re-created) by the Source addition module after source changes. Open or select the Post-processing module from the Windows menu. To examine the basic contents of an ECH-file select Utilities|Information|CATT ECH-file (or the ECH-info toolbar button). In addition to what is displayed, all early reflections are in the file with incidence angle and octave-band information. The late part is recreated in the module (statistically created based on a simplified room shape and actual incidence angle distributions). Select General settings and 44100 Hz sample-rate and binaural receiver. Select an impulse response length corresponding to the predicted reverberation time (say 1300 ms giving an IR size of 1.3 x 44100 = 57330 samples). The headphone and HRTFs used must be selected from File|Preferences or from the Prefs button. With limited versions only the "Generic" eq. (GENERIC_V8_PLAIN_44.DAT) can be chosen and only HRTFs based on a solid sphere (SPHERE_PLAIN_44.DAT). Check the settings for Output files (see to that Lake SIM-file format is selected since they will be used by the convolver in the last step). Select File|Save Settings and enter e.g. TUTORIAL.PST in CATTDATA-folder MODELS\TUTORIAL. Click Save and run and select CATTDATA-folder MODELS\TUTORIAL\OUT\E_A0_05.ECH. The module then starts to process the early part, builds the late part reverb and then finally it creates the late part. The plot-file E_A0_05_I.PLT, see Fig. 11.12, shows the early part of the created binaural impulse response, and E_A0_05_D.PLT shows the full decay for each octave-band, see Theory for details. The detailed look of the binaural impulse response depends on the HRTFs and headphone eq filter selected. Note that since head-direction “stage” was selected at prediction, receiver 05 is looking straight ahead and not towards the source and will thus have a stronger direct sound response for the right ear. To view and analyze the impulse response (via a Frequency response and a Spectrogram) contained in a SIM-file elect Utilities|Information|Lake SIM-file (or the SIM-info toolbar button). Next step is to convolve anechoically recorded material with the synthesized binaural impulse response. Click Convolution and select the anechoic 44100 Hz mono WAV-file of your choice (e.g. MTB_44_AN.WAV with a male talker or DEMO_22_AN.WAV for limited versions since IRs were then downsampled to 22050 Hz) and the just created left/right SIM-files: CATTDATAfolder MODELS\TUTORIAL\OUT\E_A0_05_L.SIM and E_A0_05_R.SIM. Finally enter a desired name for the resulting stereo WAV-file that will hold the binaural simulation e.g. CATTDATA-folder MODELS\TUTORIAL\OUT\MTB_A0_05.WAV. Play the resulting WAV-file using the CATT WAV-file player (Utilities menu), attach it to a list plot-file in the Plot-file viewer module, use the Windows MediaPlayer, load the file into a WAV editor (included with most soundcards), or use any other of the many programs that can play WAV-files.

288

11. TUTORIAL % 100

LEFT

EAR

50 0 -50 -100

0 23

50 73

100 123

150 173

200 223

ms

250 273

% 100

RIGHT EAR

50 0 -50 -100 0 23

Fig. 11.12

50 73

100 123

150 173

200 223

ms

250 273

E_A0_05_I.PLT from Post-processing

Of course, being a hall design out of the blue, the acoustics is perhaps not very good. Note that the reverberation time is nearly constant over the octaves, not dropping much at 4 kHz, and it contributes to a bright sound. Also note that the directivity used is OMNI, while a talker has a higher directivity at high frequencies and thereby in reality sends out less total power than when OMNI is used. Also this contributes to a bright sound, especially from 4k and up. Going back setting the directivity to that of a small loudspeaker makes it sound much more natural. If experiments with e.g. the directivity sounds interesting, the whole process can be repeated automatically using the sequence processing module, see next section.

11.6 Sequence processing To repeat the prediction and post-processing/convolution steps in section 11.3 and 11.5 after e.g. a raised ceiling or a directivity change, the Sequence processing module can be used together with previously created settings-files. First edit the geometry or create a new source directivity (using the Directivity module) and change the source in SRC.LOC. If desired, other parameters can of course also be changed. Then open or select the Sequence processing module from the Windows menu. All file-names below are in the CATTDATA-folder MODELS\TUTORIAL folder but must be entered with full actual path included. Click Prediction and select the settings-file CALC.PRD detailed calculation selected. Press OK.

assuming

it

has

Full

Click Post-processing and select the settings-file TUTORIAL.PST and the echogram-file ECHfile E_A0_05.ECH. Press OK. Select Convolution and select the anechoic WAV-file of your choice (e.g. MTB_44_AN.WAV), binaural impulse response SIM-files E_A0_05_L.SIM and E_A0_05_R.SIM in CATTDATAfolder MODELS\TUTORIAL and the desired name for the resulting stereo WAV-file

289

11. TUTORIAL containing the binaural simulation (e.g. CATTDATA-folder MTB_A0_05.WAV). Press OK.

MODELS\TUTORIAL\OUT\

The sequence list-box now displays the three steps to be performed. Select File|Save Sequence As and enter TUTORIAL.SEQ. If you have MTB_A0_05.WAV loaded in the MediaPlayer, it must be closed since the convolver can't save to an opened and locked file. Click Save and Run and take a short coffee break. After the break there should be a new WAV-file to listen to. You have now gone through all major parts of CATT-Acoustic. Additional important parts are e.g. multiple source auralizations and calibration enabled by the Post-processing module Utilities|SIM-file Addition and |WAV-file Addition (full version) and |Calibration menus.

290

12. REFERENCES

12. REFERENCES [Ambisonic]

see [B-format]

[ANSI S12.2-1995] CRITERIA FOR EVALUATING ROOM NOISE [Beranek]

“Acoustics”, L L Beranek American Institute of Physics, Inc. 1986

[Barron-81]

M Barron, A H Marshall, “Spatial Impression Due to Early Lateral Reflections in Concert Halls: the Derivation of a Physical Measure” J. of Sound and Vibration, 72, 211- 232, 1981

[Barron-88]

M Barron, L-J Lee, “Energy relations in concert auditoriums. I” J. of the Acoustical Society of America, 84(2), 618-628, 1988

[Borwick]

“Sound Recording Practice” John Borwick (ed.) (Ch. 6 "Microphones" by John Borwick)1987 Oxford University Press, Oxford, New York, USA

[Bradley-96]

J S Bradley, “An international comparison of room acoustics measurement systems” Internal Report IRC-IR-714, National Research Council Canada

[Bradley-96:2]

"Predictors of Speech Intelligibility in Rooms", J.S. Bradely, JASA 80(3), Sept 1996, pp 837-845

[B-format]

M A Gerzon, “Ambisonics in multichannel broadcasting and video” J. Audio Eng. Soc. 33(11), 859-71, 1985

[Chalmers]

Chalmers Room Acoustics Group Department of Applied Acoustics, Chalmers University of Technology SE-412 96 Gothenburg, Sweden

[CLF]

The Common Loudspeaker Format, see www.clfgroup.org. From this page a free CLFViewer and a free CLF Authoring kit can be downloaded. It also included links to loudspeaker manufactures offering CLF format loudspeaker data.

[Cooper-89]

D H Cooper, J L Bauck, “Prospects for Transaural Recording” J. Audio Eng. Soc. 37(1/2), 3-19, 1989

[Cremer]

“Principles and Application of Room Acoustics”, Volume 1 L Cremer, H A Müller, T Schultz Applied Science Publishing. In many cases a reference to [Cremer] is actually a reference to other authors. It is however much easier to refer to this excellent book than to each separate paper. Those of you who want more details can then dig into the references of [Cremer].

[Cremer-2]

[Cremer], Volume 2

[Cremer-89]

L Cremer, “Early Lateral Reflections in Some Modern Concert Halls” J. of the Acoustical Society of America, 85(3), 1989

12. REFERENCES [Dalenbäck-92]

B-I Dalenbäck, “Room Acoustic Prediction and Auralization Based on an Extended Image Source Model” Licentiate thesis, Report F92-03, [Chalmers]

[Dalenbäck-95a]

B-I Dalenbäck, “The Importance of Diffuse Reflection in Computerized Room Acoustic Prediction and Auralization” Proc. Inst. of Acoust. 17, 27-34, 1995

[Dalenbäck-95b]

B-I Dalenbäck, “A New Model for Room Acoustic Prediction and Auralization” Ph. D. thesis, Report F95-05, [Chalmers]

[Dietsch&Kraak-86] L. Dietsch, W. Kraak, “An Objective Criterion for the Measurement of Echo Disturbance during Presentation of Music and Speech" (in German) Acustica 60, pp. 205-216, 1986 [Gade-92]

A C Gade Practical Aspects of Room Acoustics Measurements on Orchestra Platforms 14th ICA, Paper F3-5, Beijing, 1992

[Hodgson-91]

M Hodgson, “Evidence of diffuse reflections in rooms” J. of The Acoustical Society of America, 89(2), 765-71, 1991

[Houtgast-80]

T Houtgast, H J M Steeneken, R Plomp, “Predicting Speech Intelligibility in Rooms from the Modulation Transfer Function. I. General Room Acoustics” Acustica 46, 60-72,1980

[IEC 2068-16]

2nd ed. 2003-05 Sound system equipment. Part 16: Objective rating of speech intelligibility by speech transmission index. (STI)

[Ingaard]

http://www.ingard.com

[ISO]

IS0 3382, Building Acoustics

[ISO-scatt]

ISO/DIS 17497, draft 2001-Sept-19, “Acoustics - Measurement of sound scattering properties of surfaces - Part 1: Measurement of the randomincidence scattering coefficient in a reverberation room”.

[Jacobsen-86]

T Jacobsen, “The RASTI method for objective rating of speech intelligibility” 12th ICA, paper E10-1, 1986

[Kleiner-93]

M Kleiner, B-I.Dalenbäck, P Svensson Auralization - an Overview, J. Audio Eng. Soc. 41(11), 861-875 (1993)

[Kotorynski-90]

K Kotorynski, “Digital Binaural/Stereo Conversion and Crosstalk Cancelling” 89th Conv. AES, Los Angeles 1990, preprint 2949

[Kuttruff]

“Room Acoustics”, H Kuttruff Applied Science Publishers Ltd., London, UK

12. REFERENCES

[Kuttruff-80]

H Kuttruff, “Zur Abhängigkeit des Raumnachhalls von der Wanddiffusität und der Raumform” Acustica 45, 246-55, 1980

[Lake]

Lake DSP is now part of Dolby Laboratories, and Lake no longer supports the Huron or CP4 platforms.

[MLSSA]

DRA Laboratories USA URL: http://www.mlssa.com

[MATLAB]

The MathWorks Inc. 21 Eliot Street, South Natick MA 01760, USA URL: http://www.mathworks.com

[O'Keefe]

URL: www.aercoustics.com/papers/ioa95/ioa95_1.htm

[Ognedal]

The AbsTex software, www.ognedal.com

[Oppenheimer]

“Digital Signal Processing “, Alan V. Oppenheimer, Ronald W Schafer Prentice-Hall Inc., Englewood Cliffs, New Jersey, USA

[Pickover]

“Chaos, Pattern and Beauty “, C. A. Pickover St. Martin's Press NY 1990 ISBN 0-312-04123-3

[Roy&Gimbott-93]

A Roy, F Gimbott, “Automatic Echo Determination in Measured Room Impulse Responses” 94th AES Conv., Berlin March 1993, preprint 3573.

[Steeneken-67]

H J M Steeneken, T Houtgast, “A physical method for measuring speech transmission quality” JASA 67(1), pp.318-326

[Toole-90]

F E.Toole , "Loudspeakers and Room for Stereophonic Reproduction", th 8 International AES conference, 1990 May 3-6 (paper available from www.aes.org)

[WinFlag]

URL: www.winmls.com

12. REFERENCES

APPENDIX A

APPENDIX A - TUTORIAL FILES Complete and final GEO-, source-, and receiver-files for Section 11. Tutorial. Master geo-file MASTFULL.GEO. ;MASTFULL.GEO ;PROJECT:CATT-Acoustic tutorial for v8.0 GLOBAL step = 6 ;STEP 6 - side wall reflectors INCLUDE reflect ;general materials defined here or from library ABS wood = <12 10 8 7 5 4> {213 220 160} ABS carpet = < 10 8 6 4 3 2 > {98 209 44}

;STEP 1: audience and floor surfaces ABS audience = <35 50 55 70 85 80> L <30 40 50 60 70 80> {128 128 0}

ABS floorabs = carpet ;STEP 2: ceiling, backwall, sidewalls ABS ceilabs = wood ABS wallabs = wood ;STEP 3: entrance door ABS doorabs = wood ;STEP 4: prosc. and stage ABS sfloorab = wood ABS swallabs = wood ;STEP 1: floor, audience GLOBAL ah = 1.0 ; audience height GLOBAL rh = 0.3 ; receiver height above audience plane GLOBAL aww = 1.0 ; audience sidewalk widths GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL

fsd fsw sw pw w ew fld flt wl1 wl2 hd

= = = = = = = = = = =

1 6 fsw 2 16 1.4 3 1 6 7 fld

; ; + ; ; ; ; ; ; ; +

front-stage depth front-stage width 2 ; stage width prosc. width max hall width entrance width flank depth floor tilt first side wall length second side wall length wl1 + wl2 ; hall depth from prosc

;STEP 2: ceiling, backwall, sidewalls GLOBAL h = 8 ; hall height ;STEP 3: entrance door GLOBAL e = 0.2 ; edge width entrance door GLOBAL dh = 2 ; entrance door height ;STEP 4: prosc. and stage

APPENDIX A GLOBAL GLOBAL GLOBAL GLOBAL GLOBAL

sh sd sbw sfh sbh

= = = = =

1 ; 5 ; 4 ; h-2 ; sfh-1

MIRROR 100 200

stage height stage depth stage back width stage front height ; stage back height

; corner (100) and plane (200) offsets for mirrored part

CORNERS ;-- STEP 1 -----------------------------------------------;floor corners 1 -fsw/2 fsd 2 -sw/2 0 3 x(2)-pw 0 4 -w/2 fld 7 -w/2+ew hd 8 0 y(4) 5 x(4) fld+wl1 6 x(7) y(5)

0 0 0 0 flt 0 ; help corner for lock lock(4 7 8) lock(4 7 8)

;audience surface ;20-21 help corners to lock audience corners at ah above floor 20 x(8) y(8) z(8)+ah 21 x(4) y(4) z(4)+ah 22 x(7) y(7) z(7)+ah ;23-25 help corners to lock receiver positions at rh above audience ;surface (in REC.LOC) 23 x(20) y(20) z(20)+rh 24 x(21) y(21) z(21)+rh 25 x(22) y(22) z(22)+rh ;audience upper corners 30 x(4)+aww y(4) lock(20 31 x(4)+aww y(5)-aww lock(20 32 x(7)+aww y(5)-aww lock(20 33 x(7)+aww y(7)-aww lock(20 ;audience lower corners 40 x(30) y(30) lock(4 7 41 x(31) y(31) lock(4 7 42 x(32) y(32) lock(4 7 43 x(33) y(33) lock(4 7

21 21 21 21

22) 22) 22) 22)

8) 8) 8) 8)

;-- STEP 2 -----------------------------------------------;ceiling 52 x(2) y(2) 53 x(3) y(3) 54 x(4) y(4) 55 x(5) y(5) 56 x(6) y(6) 57 x(7) y(7)

h h h h h h

APPENDIX A ;-- STEP 3 -----------------------------------------------;entrance 60 x(5)+e 61 x(5)+e 62 x(6)-e 63 x(6)-e

door y(5) y(5) y(5) y(5)

z(5) z(5)+dh z(5)+dh z(5)

;-- STEP 4 -----------------------------------------------;prosc. and stage 74 x(2) y(2) sh 75 x(1) y(1) sh 76 -sbw/2 -sd sh 77 x(76) y(76) sh+sbh 78 x(2) y(2) sh+sfh

PLANES ;STEP 1: audience and floor planes -----------------------;audience surfaces [1 audience surface / 33 32 31 30 130 131 132 133 / audience ] [2 audience front / 140 130 30 40 / audience ] [3 audience rear / 143 43 33 133 / audience ] [4 audience front side / 30 31 41 40 / audience ] [5 audience middle side / 42 41 31 32 / audience ] [6 audience rear side / 43 42 32 33s / audience ] ;floor surfaces [7 front floor / 101 102 103 104 140 40 4 3 2 1 / floorabs ] [8 sloped side floor / 40 41 42 43 7 6 5 4 / floorabs ] [9 sloped rear floor / 107 7 43 143 / floorabs ] ;STEP 2: ceiling, backwall, sidewalls --------------------[20 [21 [22 [23 [24

ceiling / 52 53 54 55 56 57 157 156 155 154 153 152 / ceilabs ] back wall / 107 157 57 7 / wallabs ] side wall front / 5 55 54 4 / wallabs ] side wall back / 56 6 7 57 / wallabs ] front flank wall / 54 53 3 4 / wallabs ]

;STEP 3: entrance door -----------------------------------[30 entrance wall / 5 6 56 55 / (door / 63 62 61 60 / doorabs) (wall / 5 6 56 55 / wallabs) ] ;STEP 4: prosc. and stage --------------------------------[40 [41 [42 [43 [44 [45 [46 [47

prosc. walls stage side edge stage front edge stage floor stage ceiling stage back wall stage side wall prosc. top

/ / / / / / / /

2 3 53 52 78 74 / wallabs ] 2 74 75 1 / swallabs ] 1 75 175 101 / swallabs ] 176 174 175 75 74 76 / sfloorab ] 77 78 178 177 / swallabs ] 76 77 177 176 / swallabs ] 76 74 78 77 / swallabs ] 78 52 152 178 / swallabs ]

APPENDIX A Geo-file REFLECT.GEO. ;REFLECT.GEO ;PROJECT:CATT-Acoustic tutorial for v8.0 ; STEP 6 side-wall reflectors ABS backabs = <20 30 40 50 60 60> {64 128 128} ABS frontabs = <12 10 8 7 6 5> {0 0 128} LOCAL r_w LOCAL r_h

= wl1/5 = h-3

LOCAL tx = w/2 - 0.5 LOCAL ty = r_w LOCAL tz = 2.75 LOCAL ry = -10 LOCAL rz = 15 OBJECT ROTATE TRANSLATE

0 tx

OFFSETPL 100 OFFSETCO 80 MIRROR 100 200 COPY 5 2 0 0 0 COPY 10 4 0 0 0 COPY 15 6 0 0 0

ry fld+ty

rz tz

tx fld+2*ty tz tx fld+3*ty tz tx fld+4*ty tz

0 ry rz 0 ry rz 0 ry rz

CORNERS 0 1 2 3

0 -r_w/2 0 0 -r_w/2 r_h 0 r_w/2 r_h 0 r_w/2 0

PLANES ;* = auto edge [0 refl front ;the back MUST [1 refl back

diffusion + enable in General settings / 0 1 2 3 / wood* ] be defined too \ 0 1 2 3 \ backabs* ]

APPENDIX A Source-file SRC.LOC. ;SRC.LOC ;PROJECT:CATT-Acoustic tutorial v8.0

SOURCEDEFS IF step < 5 THEN A0 0.0 3.0 2.0 OMNI 0.0 10.0 3.0 Lp1m_a = <70 73 76 79 82 85> B0 1.0 2.0 2.0 OMNI 0.0 10.0 3.0 Lp1m_a = <70 73 76 79 82 85> ELSE A0 -2.0 fsd-0.5 sh+1.7 OMNI -2.0 fsd+2 sh+1.7 Lp1m_a = <70 73 76 79 82 85> B0 1.0 fsd-2.0 sh+1.7 OMNI Lp1m_a = <70 73 76 79 82 85> ENDIF

1.0 fsd+2 sh+1.7

APPENDIX A Receiver-file REC.LOC. ;REC.LOC ;PROJECT:CATT-Acoustic tutorial v8.0 IF step >= 5 THEN LOCAL dx = 2.2 LOCAL dy = 3 LOCAL y0 = fld+1 ENDIF

RECEIVERS IF step < 5 THEN 1 0.0 10.0 3.0 ELSE 1 -2*dx y0 2 -1*dx y0 3 0*dx y0 4 1*dx y0 5 2*dx y0 11 -2*dx y0+dy 12 -1*dx y0+dy 13 0*dx y0+dy 14 1*dx y0+dy 15 2*dx y0+dy 21 -2*dx y0+2*dy 22 -1*dx y0+2*dy 23 0*dx y0+2*dy 24 1*dx y0+2*dy 25 2*dx y0+2*dy 31 -2*dx y0+3*dy 32 -1*dx y0+3*dy 33 0*dx y0+3*dy 34 1*dx y0+3*dy 35 2*dx y0+3*dy ENDIF

lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23 lock(23

24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24

25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25) 25)

INDEX

#C, 83

ACLIB.DLL, 3

#DIVIDED, 267

acoustic environment, 47

#G, 84, 116

adapt rays, 55

#I, 80

addendum, 3, 19

#L, 85

ADD-file, 187, 188, 286

%, 84

AI, 21

( ), 94

aim(H,V), 111

..., 76

aiming, 51

;, 76

air

[ ], 93 |, 76 *, 94 24-bit, 15 32-bit, 15 3D balloons, 175 3D extents, 174

absorption, 47, 132, 173 density, 47 humidity, 47 Ambisonic, 227, 291 AND, 77 anechoic WAV info file, 215 WAV-file, 13

3D text, 102

ANECHOIC, 4, 13, 28, 198, 235, 237

3D-view, 50

angle arguments, 84

5-Ch Binaural Setup, 221

animation, 56, 64

5-channel, 228, 229, 247

AniScape, 227, 231

5-channel simulation, 209

array elements, 169 ARRAY0, 169

ABS, 86

arrays, 144, 169

abs(), 84, 90

Articulation Index, 21

ABS_ss.PLT, 67, 128

asterisk "*", 94

ABS1, 88

audience, 58, 148, 275

AbsC, 137, 138

audience area, 49

AbsCg, 137, 138

audience area mapping, 10, 20, 53, 132, 133, 146, 284

ABSLIBS, 3, 26, 153

absolute corner id:s, 92 Absolute Threshold, 21 absorption, 86 absorption changes, 148 absorption coefficient histograms, 67 absorption/diffusion library, 153 AbsTex, 293 ACAD, 5, 268

accurate definition of plane corners, 79

audience area mapping results, 123 audience edges, 104 audience modeling, 103 audience surfaces, 133 auralization, 135, 288, 292 Auto number, 63 Auto number mapping, 55 Auto time, 64 Auto time mapping, 55 AutoCAD, 38, 94, 103

INDEX 3DFACE, 265, 268 3DMESH, 268 DDATTE, 270 interface, 5, 27, 31, 157, 265 MOVE, 270 STRETCH, 270 AutoLISP, 5, 265, 268 APPLOAD, 268 CATTSCALE, 269 CATT_FIX, 267, 268 CATT_MAIN, 268 CATT_TMP, 268 CHDIRECT, 270 CHSRCAIM, 269 CHSRCPOS, 269 CONV, 268 DELSHDIRECT, 270 EXP, 269 GET, 269 HIDEDIV, 267, 270 IMP, 270 LOAD, 268 MARKFIX, 268, 270 PUT, 269 REC, 269 SETCATTSCALE, 269 SHDIRECT, 270 SHDIV, 270 SHR, 270 SRC, 269 UNMARKFIX, 270 DRAW, 269

Beranek L., 291 B-format, 227, 229, 246, 291 1st order, 227, 246 2nd order, 227, 246 3rd order, 227, 246 KLMNOPQ, 227 RSTUV, 227 WXYZ, 227, 246 binaural, 226, 229, 245 binaural down-mix, 223 binaural post-processing, 288 bitmap-file, 29 blank lines, 76 Blumlein, 226 Borwick J., 291 Bradley J. S., 291

BREAK, 78 C-50, 15 C-80, 138, 193 CAD-file, 24 CAD-file format, 266 CAG-file, 234 calculation recommendations, 146 calibration, 243

automatic edge diffusion, 136

calibration count, 202

automatic play, 182

camera, 50, 186

aux delay, 205

camera position, 19

A-weighted, 58

cardioid, 228

AWI-file, 112, 202, 215

categories, 155

background calculation, 23

category, 154, 158

background noise, 47, 57, 116, 140, 191, 193

CATT, 34

residual, 57 backup, 42 backward integrated, 66, 143 balconies, 133 Barron M., 291 batch processing, 253, 289 batch run, 23 Bauck J. L., 291 beam-steering, 169

CATT 3D-viewer, 2, 9, 27, 51, 122 CATT play-list, 203 CATT32.EXE, 2, 29

CATTDATA-folder, 1 CATTED32.EXE, 2, 263

CATT-Edit, 263 CATT-Edit font, 263 CATT-folder, 1, 2 CATTOGL32.CHM, 51 CATTOGL32.EXE, 2

INDEX CATTPLT32.EXE, 2, 179

CATT-registered files, 24 CATT-SD0, 26, 162 CATT-SD1, 26, 165 CATT-SD2, 161 CATT-VRoom, 35 CATT-Walker, 3, 25, 68, 118, 129, 202, 209, 227, 233, 259 CF1, 171 CF2, 171 Chalmers, 34, 291 character case, 76 character set, 24, 263

CIRC (marker), 101 Clarity, 138 CLF, 40, 291 CLFViewer, 172 closed model, 53 Closest, 59 CMB-file, 53, 62, 68, 122, 127, 131, 187, 188, 286

corners previously defined, 91 single-connected, 52, 103 used, 52

CORNERS, 79, 89 cos(), 84 2

cos weighted, 127 create a GEO-file, 41 create project, 273 Create Project, 6, 41 Cremer L., 291 crosstalk cancellation, 213, 248, 261 crosstalk cancellation filters, 214 cube(), 84, 90 cut(), 91

CWI-file, 129, 209, 233 D-50, 138, 193 Dalenbäck B.-I., 292

DASH (marker), 100 data flow, 25 data output, 32

COINFO.PLT, 49, 105, 122, 277

DATA32.DLL, 3

collection files, 26

DDI, 3, 32, 144, 159, 172

collections, 3

debug, 51

COLORED.PLT, 49, 121

DEBUG.TXT, 51, 122, 150

comments, 76

decimal character, 24, 76, 275

connecting to adjacent planes, 103

default diffusion, 48, 88

constants, 147

delay, 59, 114

control rooms, 32

Deutlichkeit, 138

convert

diffraction, 135

to ANSI, 263 convolution, 197, 230, 234, 240, 259 Cooper D. H., 291 coordinate system, 31 copy 3D view, 181 a GEO-file, 41 bitmap, 181 Meta-file, 180

COPY, 83, 148, 282 copyright, 4, 19, 219

diffuse reflection, 14, 47, 138, 147, 283, 292 estimate, 87 frequency dependent, 134 order, 61, 126 diffusion, 86 direct sound, 13, 63, 133, 135 directives, 79, 80 directivity, 144, 159 balloon, 17 collections, 159

INDEX contours, 20, 49, 51, 59, 122, 175, 178 factor, 178 index, 168, 178 polar charts, 175

DISC (marker), 101 disproportionate rooms, 135 DLL Directivity Interface, 3, 32, 159, 172 Dolby Surround, 223 doppler, 236 double-sided planes, 93, 98, 122 duplicate corner id:s, 52 corners, 52 plane ids, 52 planes, 52 DXF, 27, 265 DXF2GEO, 2, 31, 265 DXF-file, 265 Dynamic Link Libraries, 3 E_rr.TXT, 193 E_rr_INT.PLT, 193 E_rrX.TXT, 194 E_ss_rr.ADD, 62, 68, 127, 131 E_ss_rr.ECH, 62, 68, 127, 131, 191, 192, 197 E_ss_rr.TXT, 64, 129 E_ss_rr_INT.PLT, 66, 128 E_ss_rr_oct.PLT, 65, 127

ear integration, 143 Early Decay-Time, 140 early part, 134 early part detailed ISM, 10, 60, 132, 133, 146, 284 early part detailed ISM results, 125 early part response, 231 ECH-file, 187, 198, 224 echo, 149 echo-disturbance, 66, 128, 141 echoes, 149 echogram, 143 echogram ISM, 61 echogram smoothing, 16

echograms, 65 edge contours, 49, 122 diffusion, 48, 137 penetrating, 52 EDT, 140 EKgrad, 66, 128, 141, 193 electro-acoustical source, 174 e-mail, 36 Energy relations, 291 eq/gain, 226 Err_nn_oct.PLT, 192 ERRORS32.DLL, 3 estimate(), 87 EXAMPLE, 5

Example model, 105

EXCLUDE, 82 exp(), 84

export of absorption data, 154 of bitmap-file, 180 of maps for spreadsheet, 180 of Meta-file, 179 to AutoCAD, 38, 265, 266, 270 to Merged GEO, 41 to MRK, 40, 88 to VRML, 38 expression, 84, 89 External CATT Tool, 11 Eyring, 137 Eyring RT, 69 EyrT, 137 EyrTg, 137 F1, 19, 24 FALSE, 77, 84

field of view, 50 figure-of-8, 228 file calibration, 204 file conversion, 210 file scaling, 204 file-name syntax, 37 filter, 65

INDEX rectangular, 65 exponential, 65 triangular, 65 first-order surface id:s, 62 floating point values, 76 font-size, 13 forward integrated echogram, 143 fractional delay, 226 free path histograms, 67 frequency range, 32 frequency response, 200

FROMBACK, 82, 94 FROMFRONT, 82, 94 frontal normalization, 218 full detailed calculation, 10, 63, 132, 134, 146, 285

Head-Related Transfer Functions, 4 HeadScape, 15, 28, 232, 233 head-tracking, 28 help, 24, 264 help button, 19 help-file, 3 hidden lines, 88 hidden options, 20, 126 HIDDENOPTIONS.TXT, 20

Hilbert transform, 244 history file, 20 history plot, 15 HITS_ss.PLT, 67, 128

Hodgson M., 292 holes, 94

full detailed calculation results, 127

Houtgast T., 292

functions, 30

HRTF, 4, 203, 217, 218, 245

G (strength), 139 Gade, A.C., 292 Gain, 205 Gain_a, 177, 190

GEO-file sections, 79 GEO-file tree, 27, 42, 80 geometrical acoustics, 32, 71 geometrical errors, 150 geometry view and check results, 121 Gerzon M. A., 291

GETGLOBAL, 79, 85, 116 GETLOCAL, 9, 79, 85, 110 GLOBAL, 6, 79, 84, 116 graph color, 16 H_ss_rr_oct.PLT, 62, 126

hardware locks, 1 hardware requirements, 29 head-direction, 46, 115, 116, 139, 246 headphone, 203, 211, 221, 243, 246 eq, 261 equalization, 4 filters, 13 HEADPHONES, 4, 13, 28, 198, 220

interpolation, 245 library, 13 normalization, 219 HRTFS, 4, 13, 29, 198

HTML, 40 HTR-file, 232 HURON1, 232

hyper-cardioid, 228 I_ss_rr.TXT, 126 I_ss_rr_oct.PLT, 61, 126

IACC, 213, 226, 248 IEC 2068-16, 292

IF-THEN-ELSE statements, 77 Image Shift, 21 Image Source Model, 133 image source space, 62 Image Spreading, 21 imperial units, 31 import of absorption data, 153 of directivity, 162, 165 impulse response, 197, 242 display, 199

INDEX inaccurate corners, 52 incidence angle distribution, 247

INCLUDE, 45, 78, 80, 148 input folder, 44 input/output files overview, 25 installation, 1 AutoCAD interface, 268 Inter-Aural Cross Correlation Coefficient, 213, 248

expiry date, 19 file, 3

LINE (marker), 100 list-file, 181 LMB, 184

LOAD (marker), 102 LOCAL, 6, 79, 85, 110, 116 lock(), 91, 276

locking corners, 91

Inter-Aural Cross Correlation Function, 248

log(), 84

interference, 58, 189

log10(), 84

interpolated, 168

long filenames, 24

INTERPOLATED0, 168

loops

IR, 197 IS0 3382, 292 ISM, 60, 133 Jacobsen T., 292 Kotorynski K., 292

for planes, 96 loop(), 79, 92 notes, 119 override ids, 117 recloop(), 117, 264 recloop2(), 117, 264 recwalk(), 117, 118, 264

Kuttruff H., 292

lost rays, 152

Lake SIM-file, 199, 229

loudspeaker

Lake Technology, 293 Lambert, 86 Lambert's law, 133 late part response, 232 late part reverberation, 247 Lateral energy fraction, 139 lateral reflections, 291 Latest Results List, 41 law of the first wavefront, 188 leaks, 129 LEAKS_ss.PLT, 129

Lee L.-J., 291 lens angle, 50 LF, 60, 139, 193 LF_oct.PLT, 123 LF_oct_n_3D.PLT, 124

LFC, 139, 193 LFDIR_oct.PLT, 60, 124 LFDIR_oct_n_3D.PLT, 124

license, 19

correction, 214 listening, 213, 248 Lp_max, 112, 113 Lp_nominal, 112 Lp_pink Lpa1k, 112 Lp_pink Lpea1k, 113, 114 Lp_voice_IECfemale, 112 Lp_voice_IECmale, 112 Lp_voice_loud, 112 Lp_voice_normal, 112 Lp_voice_raised, 112 Lp_white Lpa1k, 112 Lp_white Lpea1k, 113 Lp1m_a, 7, 111, 190 Lp1m_ea, 190

MakeWalkerBinDec, 2 manifest-file, 5 map height, 58, 123 map step, 58 mapping

INDEX scale, 15, 60, 183 time intervals, 59, 60 MAPRAYFACTOR, 55, 111

margins, 180 marker elements, 31 marker files, 26 marker-file, 40, 99, 102, 161, 174 Markers, 99, 264

MARKERS, 79 Marshall A. H., 291 MASTER.GEO, 6

master-file, 45, 79, 275 MASTFULL.GEO, 273, 295 MASTSTEP.GEO, 273

MATLAB, 210, 211, 229, 230, 293 Max SPL, 174 maximum file size, 263 MDI, 29 mean absorption coefficient, 138 mean free path, 138 menus, 11 merge planes, 41, 265 Meta-file, 29, 179 METER32.DLL, 3

microphone, 226 microphones, 246, 291 mini tutorial, 6 minimum phase, 244 mirror, 30, 79

MIRROR, 82, 148, 271, 276 mixing shape, 138 MLSSA, 16, 210, 211, 229, 230, 293 MLTVLV32.CHM, 5 MLTVLV32.EXE, 5

modeling, 273 notes, 103 tips, 271 mono, 229, 246 morphing, 236 moving around, 236

Müller H. A., 291 multiple directives, 77 multiple source auralization, 187, 197, 205, 207 MultiVolver, 5, 15, 209, 223, 228, 236, 247, 249 MULTPLT.TXT, 192

natural source, 173 negative numbers, 77

NOHIDELINES, 88 noise mapping, 47, 56 NOISEMAP_oct.PLT, 124 NOISEMAP_oct_3D.PLT, 124

normalization, 220 diffuse, 219 frontal, 219 plain, 219 other, 219 normalized parameters, 64, 140 normalized values, 68

NOROOM, 5, 40, 88 NOT, 77 NTREND_ss.PLT, 68, 129

number of characters in names, 157 characters in names, 77 corners, 30 planes, 30 rays, 127, 135 rays full detailed, 63 rays mapping, 55, 123 receivers, 32 sources, 32 sub-divisions, 30 surface properties, 30, 31

OBJECT, 79, 80, 282 octave-band, 37 offset, 99

OFFSETCO, 81, 89, 282 OFFSETPL, 81, 282 OGLSWITCH32.DLL, 3

omni, 228 OMNI, 111, 114, 190

omni-directional source, 159

INDEX Open Recent, 11

Plomp R., 292

open rooms, 135

plot-file, 179

Oppenheimer A. V., 293

plot-file control, 14, 19, 179, 182, 184

OR, 77

plot-file frame, 183, 185

origin, 31

plot-file lists, 181

ORIGIN, 81

plot-file viewer module, 29

output folder, 44

PLT….TXT, 27, 121

overlapping planes, 52, 53, 103

polar charts, 175

Page Setup, 180

POLY (marker), 99

page size, 180

POSTPLT.TXT, 28, 231

palette

post-processing, 62, 68, 191, 197, 259

for color mapping, 184 for shaded 3D plots, 184 PARAM.TXT, 194 param_oct_3D.PLT, 124

PARAM_ss.TXT, 131 PARAM_ssX.TXT, 131 paramDIR_oct_3D.PLT, 124

parameter map, 67 trend, 68 parametrical modeling, 266 PARAMn_ss.PLT, 67, 129 PARDIRn_oct.PLT, 59 PARn_oct.PLT, 59, 124 PARnDIR_oct.PLT, 124 PATH_ss.PLT, 67, 128

phase, 58, 123, 189 Phone/fax, 36 PI, 21 Pickover C. A., 293 plane corner entering order, 49, 93, 94 definitions, 93 ids used, 52 sub-division, 30, 94, 95, 278 use, 46 plane-line intersection, 91

PLANES, 7, 79, 93 Play WAV-file, 182 PLINFO.PLT, 49, 122

data flow, 197 PP, 197 PPW-file, 24, 235 prediction, 257 Prediction module, 26, 37 prediction output-files, 121 Preferences, 11 printer, 180 programname.CHM, 3 programname.HLP, 3 programname.PDF, 3

project name, 44 PROJECT.TXT, 6

protect directivity data, 169 PST-file, 197 PureVerb, 5, 227, 228 PUREVERB32.CHM, 5 PUREVERB32.EXE, 5 R_ss_rr_INT.PLT, 128 R_ss_rr_oct.PLT, 66

Randomized Tail-corrected Cone-tracing, 63, 134 RApid Speech Transition Index, 140 RASTI, 47, 57, 58, 68, 124, 129, 140, 191, 193, 292 RASTI.PLT, 57, 124 RASTI_ss.PLT, 68, 129 RASTI_with_noise_3D.PLT, 124 RASTI_without_noise_3D.PLT, 124

INDEX ray truncation time, 55, 64, 123, 127

RTC, 63, 132, 134

ray-tracing, 133

RULE (marker), 100

REC.LOC, 6, 45

S_ss_rr_oct.PLT, 62, 127

receiver

Sabine, 137

file, 45, 115, 281 id, 37 location, 116 model, 226, 245 positions, 148

RECEIVERS, 7, 115, 116 REFLECT.GEO, 273, 298

reflection analysis, 60 density, 247 density growth, 134 history, 62 path magnitude, 243 path transfer function, 243 path transfer function phase, 244 transform-size, 225 reflectors, 148, 282 REFLINFO.PLT, 49, 122, 282

REPEAT, 255 reserved characters, 76

RETURN, 78, 150 reverberation time, 137 reversed normals, 103 planes, 52

Sabine RT, 69, 104 SabT, 137 sample-rate, 225 conversion, 212

SAY, 77 SCALE, 80, 110, 114, 116, 117, 150 scale factor, 199 scale factors, 242 scattering coefficient, 133 Schafer R. W., 293 Schröder, 128 Schröder M., 71 Schultz T., 291 SCL-file, 202, 204, 234 screensaver, 15 SD, 3, 26, 159

SD0, 159, 170, 176 SD1, 159, 171, 176 SD2, 144, 159, 172, 177 SD2DATA, 4, 160, 161, 172

sector processing, 246

RGB-color, 6, 99, 264

sensitivity, 164, 167, 174

Rise time, 127, 143

Separate Echo, 21

RMB, 184

sequence processing, 253, 289

rnd(), 84, 90

Sequence processing module, 29

rnd2(), 84, 90

SHADED.OGL, 9, 51, 122, 277

roll angle, 111

SHADED.PLT, 49, 121

room acoustic parameters, 33

SHIFT, 80

Room Acoustics, 291

shoebox, 6

room acoustics measurement systems, 291

SHOEBOX, 4

ROTATE, 80 RT, 137 RT overview, 67 RT_ss.PLT, 67, 128

RT’, 59, 138

Shuffler, 248 SI units, 31 SII, 21 SIM-file, 198, 199, 204 addition, 205, 261 filtering, 211

INDEX sin(), 84

SPL_oct_n_3D.PLT, 123

slanted ceilings, 147

SPLcont, 59

slanted floors, 147

SPLdir, 59

smoothed

SPLDIR_oct.PLT, 123

echogram, 127 responses, 65 smoothing, 61 soffit-mounted source, 149 software overview, 29

SPLDIR_oct_n_3D.PLT, 123

spline interpolation, 244 spread-sheet export, 33 programs, 131

Sone, 143

sqr(), 84, 90

Sone-scale, 16, 61

sqrt(), 84

sort order, 157

SRC.LOC, 6, 45, 110

sound roses, 66, 128

src.PLT, 192

soundcard, 29, 198

SRCINFO.PLT, 49, 121, 187, 191

source

SRCINFO3D.PLT, 49, 122

addition, 27, 62, 68, 187, 197, 286 aim, 49 combination, 187, 188 combination data, 27 directivity, 3, 26, 32 directivity module, 26 file, 45, 110, 280 id, 37 roll angle, 49 source addition module, 27

SOURCEDEFS, 7, 110 space bar, 268 SPECTRA.PLT, 194 SPECTRA_ss.PLT, 68, 129

spectrogram, 201

Stage support, 63, 142 startup.PLT, 179

statistical properties of the room, 134 statistics, 67, 70, 124 statusbar, 24 STearly, 63, 142 Steeneken H. J. M., 292, 293 stereo, 229 STI, 16, 21, 47, 57, 58, 68, 124, 140, 191, 292 STI_name.PLT, 124 STI_name_with_noise_3D.PLT, 124

specular reflection order, 61, 126

STI_name_without_noise_3D.PLT, 124

specular reflections, 63

STI_ss.PLT, 68

Speech Intelligibility, 292

STlate, 63, 142

Speech Intelligibility Index, 21

stopband suppression, 226

speech levels, 140

Strength, 139

Speech Privacy Index, 21

string constants, 30, 84

speech spectrum, 114

STRNGS32.DLL, 3

Speech Transmission Index, 16, 140

STtotal, 63, 142

speed of sound, 132

studios, 32

sphere HRTFs, 217

super-cardioid, 228

SPL, 60, 193

surface

SPL_oct.PLT, 60, 123

graph, 180 properties, 3, 13

INDEX properties module, 153

trouble-shooting, 150

Surface properties module, 26

TRUE, 77, 84

sweetspot, 250

truncation time ISM, 61

SWITCH32.DLL, 3

Ts, 193

symbolic constants, 30

tutorial, 273

symmetrical halls, 79

TUTORIAL, 5, 273

syntax

U50, 21

GEO/LOC, 264 Marker, 264 T-15, 137 T-30, 137 TAB, 76 tail sectors, 246 tan(), 84

temperature, 47 template markers, 264 source, 264 template.PST, 198 template.PRD, 37

TEXT (marker), 102 The FIReverb Suite, 5, 35 time-constant, 16, 65 of the ear, 143 TIM-file, 210 tips and tricks, 147 toolbar, 19, 186 Toole, F. E., 293 tools, 30, 79 tooltips, 19, 186 TOTABS, 86, 148, 151, 157 TOTREF, 86, 157

trademarks, 1 TRANSAURAL, 214

TRANSLATE, 80 transparency, 86, 87, 145 tree-structure, 78 TREND1,2.PLT, 194

UNCTT.EXE, 2

under-balcony positions, 135 unit, 77 user interface, 11 user-supplied information, 6 V_ss_rr_oct.PLT, 66, 128

variables, 254 vectorgram, 66, 128 view angles, 50 distance, 50 VIEW4.PLT, 48, 121

virtual reality, 198 visuals, 174, 176 vocal effort loud voice, 115 normal, 115 raised, 115 volume, 49, 71, 121, 131 calculation, 103 estimate, 53 VRML, 27, 34, 38 WalkDecim, 3 Walker.CAG, 25, 68, 129

walkthrough convolution, 35, 236 walkthrough convolver, 15, 232 script-file, 237 wall hit statistics, 67 warnings, 13 warped planes, 129 WAV-calibration, 260

TREND3.PLT, 194

WAV-file, 179, 181, 182, 185, 198, 202, 204, 207, 210, 211, 230, 231

TRENDn_ss.PLT, 68, 129

WAV-file addition, 207

INDEX WAV-file info, 112

www, 36

WAV-file player, 203, 236

x(), 91

WCS-file, 237

XOR, 77

white

XTRA32.DAT, 3

space, 76 spectrum, 149 WinFlag, 293 wire-frame, 184 WRL-file, 24, 40

XTRA32.DLL, 3

XY-stereo, 226, 246 y(), 91 z(), 91

ADDENDUM TO PRINTED MANUAL This document lists manual clarifications and all additions after printing of the v8.0g:2 manual in Aug 2007 and covers up to v8.0h:1 in Sept 2008. v8.0h:1 updated a big number of modules due to internal library changes, the actual functionality changes are listed below. The help-file (CATT32.CHM) is always up to date as are dedicated help-files such as for CATTOGL32.CHM for the 3D-Viewer (CATTOGL32.EXE). News is added from the top in each section.

Manual clarifications Page 38: VRML has since some time been replaced by the X3D format but one version of X3D is essentially VRML (minor header differences) and e.g. the http://www.octaga.com X3D player can read the exported WRL files. Page 122: Additional docs about the addition of plane GEO-file:line-number readout in the PLINFO.PLT plot-file: planes created using COPY will all have the same line number. Page 113: the Max SPL values on axis as set in the Directivity module for an electro-acoustic sources and uses as Lp1m_ea = Lp_max are not available for CLF sources ( .CF1 and .CF2). Page 64: when using the Prediction Full detailed calculation option Late part ray-trace, do not put receivers close to surfaces (such as the audience). With this option the receiver sphere size grows, as in the RTC, until it hits the closest surface and from that point ray-tracing is used and ray-tracing with a small receiver sphere will give a larger random error requiring many more rays. Page 191: the Ref. SPL spectrum selection documentation is missing, refer to the help-file. 12. References A useful reference regarding coupled rooms and the use of Late part ray-tracing in Full detailed calculation: [Summers-2005]

Jason E. Summers, Rendell R. Torres, Y. Shimizu, Bengt-Inge L. Dalenbäck Adapting a randomized beam-axis-tracing algorithm to modeling of coupled rooms via late-part ray tracing J. Acoust. Soc. Am. 118, 1491 (2005)

Prediction module v8.0h:1 Full detailed calculation and Audience area mapping speedup via parallel processing As was indicated below with v8.0g:3 Interactive RT estimation speedup it was uncertain if any further parts of CATT-Acoustic would utilize parallel processing due to the fundamental algorithm changes required but a safe transformation procedure was developed and the ray-tracing in Full detailed calculation as well as Audience area mapping ray-tracing now runs multi-threaded and will use all cores (CPUs) of multi-core PC, for a single-core PC it will be the same speed as before. The actual calculation speed depends on case ranging from no increase for a simple model to 2x for a dual-core and 4x for a quad-core and a complex model. However, multi-element SD2 arrays especially will so far not benefit from the multi-threading since the main benefit is on the ray-reflection calculations so for cases when the array calculations are significant the speedup cannot reach full potential. If, for some reason - such as running another type of calculation in parallel, making full use of all cores is not desired a Preferences

Addendum to Printed Manual for v8.0h:1 option allows limiting the ray-tracing to one core. This setting does not affect Interactive RT estimation that will always run on multiple cores being intended as an interactive type of calculation. v8.0h:1 A new prediction display “Time trace” At the bottom of the Prediction settings dialog is a new button Time trace plot-file..., showing a dialog with options on tracing rays one time segment at a time creating a 3DA PLT-file (i.e. a file with animated frames, each frame representing one time segment): Band: selects the octave-band for which the time trace holds. No of rays: the number of rays to use, 10k is a typically sufficient number for most (not very big) rooms. Max time: the max time to trace the rays (counting from the source excitation unlike with all measures that start counting when the direct sound hits a receiver). If an electro-acoustic source has a delay of say 10 ms no rays will appear from that source until after 10 ms, if a source has a delay it is indicated below the source position. Typically good values for Max time are 100-200 ms but is case dependent. Time step: the time step from frame to frame. If a small value is selected and many rays it will be more like a wave-front display but will also create a big plot-file (> 50 MB not uncommon, if the resulting file will be huge there is a warning given and the processing may even be aborted, see more below). Typical good values are 1-2 ms but is case dependent, up to 10 ms (= 3.4 m) is allowed but is hardly useful except in big models or quick initial tests. Max order: the max reflection order of the rays traced, the value depends on what is to be illustrated, selecting 0 will show only the direct sound, selecting 1 will give one wall reflection etc. Min level: to how low relative level the rays from a source are to be traced counted from the level at 1m on axis, the value depends on what you want to illustrate. This value does not include the 1/r2 spreading but only directivity, absorption, scattering and transmission losses. For example, selecting -20 dB will not show any rays at the rear of a directive speaker with < -20 dB rear suppression, or for an omni, rays that have hit two pure absorbers with α = 0.90 (10 log((1-0.9)⋅(1-0.9)) = -20 dB) will be truncated. Scale max: what SPL is represented by pure red on the color scale, this affects only what colors are used for each level and not the calculation as such. Rays with a higher level than the Scale max will be shown gray (the first few frames will typically have a very high level due to 1/r2 with r < 1m) while rays with a level ≤ the scale min will be black. For example, for a simple electro-acoustic source where 80 dB is set on the axis at 1m (Lp1m_ea or Lp1m_a + Gain_a) a Scale max of 80 dB will give red for rays at 1m on axis and then lower colors further out as 1/r2 decreases. Since it seldom is useful to see the levels very close to a source it is better to set a Scale max closer to the SPL that what will be at the audience. With distance dependent SD2 arrays it is more complicated to set an initially good value but start with few rays and Max order 0 to find a good value. When the direct rays that hit the audience are red it is a good setting of Scale max for normal use. Ray length: the Time step, n px wide choices use a ray segment length equal to the Time step, the Relative, n px wide choices use a short ray length relative to the room size (room diagonal / 200) unless that leads to a longer ray segment than the time step. The 2 and 3 pixel choices make rays more visible but the drawing a little slower, it is most useful when the number or rays have to be few to cut down the calculation time or plot-file size. Show lost rays: this can be a useful option in an actual open room / outdoor case (or if only a few model planes are used in an otherwise closed room, like e.g. reflectors and an audience), otherwise it will not matter since no rays (should) escape.

2

Addendum to Printed Manual for v8.0h:1

On: rays that escape will be visible until they hit an invisible surface just outside the model's max extents. Off: rays that never hit anything at all will not be seen at all. Show receivers: shows, for reference, the normal receivers with their ids. Write plot-file to temp folder: the plot-files can be very big (> 50 MB not uncommon) but are typically quick to create (for normal sources) so it may be better to let them be created in a Windows temporary folder so that they will be overwritten each time rather than to have several of these big files in every project output folder cluttering the disk. If the plot-file size is estimated to be > 50 MB a warning is given (indicating the estimated size) before it is created. OK: starts the processing and when done loads the plot-file for viewing, the plot-file name is TIMETRACE_oct.PLT in the output folder or a in a Windows temp folder if Write plot-file to temp folder is selected. NOTE: for big plot-files, when loading the file in the viewer and building the frame structure with lots of little ray segments can take some time during which you will see the hour-glass cursor. Less than 50 MB is quick enough though. During calculation the processing dialog indicates the estimated size as time segments are created and if the size exceeds 200 MB the calculation is aborted and lower settings have to be selected. The calculation time grows as follows: with the number of sources and rays, with the Max time, the Max order and inversely with the Time step but is mitigated by a high Min level. Normal sources are fast but SD2 arrays with many elements take much longer time since the distance dependent directivity has to be evaluated for every time step and not only at receiver hits as in normal prediction. If the calculation seems to take too long, or the file size too big (see processing dialog) Stop it and you are back at the Time trace dialog and can adjust some settings and try again. Time trace will not run multi-threaded in the first version but that would anyway not affect the longer calculation times of SD2 arrays (to evaluate the directivity of a 16-element array of CF2 or SD3 elements will take roughly 50 times longer than a simple source (as with other prediction methods, up to the second wall hit in the time trace it is evaluated every time step, from then on the far-field directivity is used). To make a display like this meaningful, randomized diffuse reflection and transmission cannot be used explicitly (the display would just show a big mess) so what you see is only the specular fraction affected by - source level - directivity - absorption - scattering - transmission - air absorption - 1/r2

(distance dependent for an SD2 array) ⋅ (1-α) at each wall hit ⋅ (1-s) at each wall hit ⋅ (1-τ) at each wall hit (if a semitransparent wall) (the distance dependent directivity of an SD2 array will give the 1/r near-field behavior)

where the scattered and transmitted fraction is considered lost. This is actually a good thing in a display like this since it allows for more clear checking of e.g. problematic back wall reflections and similar since it only the potentially echo-creating specular fraction will be seen. The Time trace plot uses the selected Geometry view/check 3D view so if Edit|Copy 3D-view to PRD the selected view will be used the next time trace. It is often best to use a parallel perspective (P) and then use the X Y Z keys to get useful views. WARNING: this type of display has been requested by some users and it is only after a very long hesitation it has been implemented. The reason is that the time trace is a source-centered display rather than a receiver-centered display and it is difficult to deduce from the display what an echogram at a certain receiver looks like, and it is the echogram that decides what the listener hears. A time trace may look great (in some sense) but there may still be bad echograms at the receivers. In this display, when rays arrive at a listener position, you have to mentally consider that the listener actually hears nothing until the direct sound arrives and then hears everything relative the arrival of that direct sound, this part is

3

Addendum to Printed Manual for v8.0h:1 not (cannot be) shown in a display like this but is included in all other prediction functions that are to be considered acoustically more relevant. For example, using a 100 ms Max time will not show any reflections at all at a receiver located 34 m or more from the source (since the direct sound itself takes 100 ms to arrive ≈ 34 m) while it will show some reflections for positions closer to the source so it is deceptive, especially in long rooms. What may be useful ways to use the time trace display are: ⋅

see overall room shape problems (like a curved rear wall but that should be rather obvious anyway)

⋅

find problematic reflections (but if the model is big it will require many rays and a high Max order)

⋅

investigating reflector coverage (probably clearest to switch off all planes but reflectors and audience and use Max order = 1)

⋅

indicate room shape problems to non-acousticans but in many ways this is a too simplistic way of looking at sound and the visual aspect of the display may take precedence over actual usefulness and facts, this display should only be a complement not the main argument.

⋅

when, in time, the sound field seems to become diffuse due to the "mixing" by the room geometry (but it is likely to be diffuse before that time due to scattering in this display only indirectly shown by weakening of the specular rays).

⋅

for sound systems, to visualize how loudspeaker delays work.

⋅

get some insight in how coupled volumes or stage houses may work (but to really see coupled cases both Max time and Max order has to be high so the plot-file will be very big and slow to work with).

⋅

teaching students new to acoustics, I think this is probably the best use. It will also give some insight in how prediction works (direct sound and specular part at least).

By varying especially Max order and Min level it is possible to emphasize certain things more clearly. It is necessary to limit the number of time segments in some way (or it would be information overload and the display useless) but it can also mean that essential reflections are never discovered, to go from what this display alone shows to what a listener actually hears is a long way so the advice is to use this display with caution. v8.0h:1 Hidden option for not calculating contours The old hidden option (AC5D6-3EC76-0A444-9B627) for not calculating any contours is operational again. This is useful with some complex DDI ARRAY0 arrays for which it may take quite some time to calculate (but they only have to be calculated once). At the same time these types of arrays seldom have the main lobe on axis so the contours are not really meaningful. v8.0g:4 Faster animation echogram drawing The Full detailed calculation animation option uses a faster curve drawing technique for the echograms so that animations will run faster. v8.0g:4 ISM ray bundle plot The Early part detailed ISM / Refl. history plot has a new option dialog including an option to also create a file that shows all rays for all orders (J_ss_rr_oct.PLT).

4

Addendum to Printed Manual for v8.0h:1

v8.0g:3 Interactive RT estimation speedup Interactive RT estimation ray-tracing now runs multi-threaded and will use all cores of a dual or quad core PC, for a single core PC it will be the same speed as before. The actual calculation speed depends on case ranging from no increase for a simple model to twice (for a dual core) and a complex model. Note: if any further main parts of the current CATT-A will be rewritten to use more than one processor is uncertain. The global ray-tracing used in Interactive RT estimation uses a much simpler algorithm than Full detailed calculation and Audience area mapping, that also have many options added over time such as for animating rays, and was reasonable and safe to rewrite for parallel processing while in most cases an algorithm has to be written with that in mind from start. Converting an old complex algorithm is difficult, will lead to new bugs, and to further complicate it, a multi-threaded application is also harder to debug. v8.0g:3 Sub-division check New plane sub-division checks that will give error messages: •

if, in a case where sub-divisions are on top of each other (typically door and window in a wall, see Fig. 2.3c page 95), the last division is not the complete surface

•

if only one sub-division is defined.

Source addition module v8.0g:4 Overall SPL eq An option where the source level and spectrum can be changed for all sources simultaneously. Typical use is with optimizing a system regarding IEC STI where the overall level affects the results both in relation to the background noise (raising the SPL will increase STI) and the absolute SPL (due to level-dependent masking, raising the SPL too much will decrease STI). In the Utilities|Show Source Combination display and text-file results the eq spectrum is indicated and shown added to all source levels. This setting affects all results as well as any created or recreated ECH-files but is to not saved when the Source addition dialog is closed. Note: this is aimed for experiments and so far no checks are performed if e.g. a calibrated DDI array after overall SPL eq get a gain > 0.0 dB.

Directivity module v8.0g:4 more decimals in SD2 polar plot distance

DLL Directivity Interface (DDI) v8.0g:4 A new DLL for d&b audio J-series curved line arrays A new DLL for d&b audio J-series (DandB_CATT). The rules for the number of possible cabinets are complex and depend on array curvature, frame angle and other details and are not taken into account, refer to the d&b audio ArrayCalc software for determining possible combinations. Be sure to read the DandB_CATT.PDF help (press the Help button after a new J-series SD2 has been selected) that contains info about how the DLL operates.

5

Addendum to Printed Manual for v8.0h:1 v8.0g:3 A new DLL for Renkus-Heinz Iconyx A new DLL for Renkus-Heinz Iconyx IC8, IC16, IC24 and IC32 DSP-controlled line arrays. This DLL is different from most other DLLs since it requires use of an external software (Renkus-Heinz BeamWare 1) to create the FIR-filters. When a new Iconyx SD2 is created a default FIR-filter set is used but for any further use BeamWare 1 is required. Be sure to read the Iconyx_CATT.PDF help (press the Help button after a new Iconyx SD2 has been selected) that contains info about how the DLL operates and where to obtain BeamWare 1. v8.0g:3 Updated DLL for Active Audio StepArray The StepArray DLL uses a new set of steering parameters and the SA250S model is added. Any old StepArray SD2-files will fail, contact Active Audio for information about how to convert old parameters.

Surface properties module -

Post-processing and CATT-Walker (full version) v8.0g:4 Faster SIM/CWI-file info IR drawing The SIM/CWI-file info dialogs optionally uses a faster curve drawing technique for the case where the number of data points is higher than the number of pixels. CATT-Walker v1.1f (with v8.0g:4) Direct B-format output CATT-Walker has Output Format options Binaural (same as previous versions) and the new B-Format. Bformat requires a 4-channel capable soundcard to be installed and selected and an ambisonic loudspeaker decoder to be used externally. For details see Walker.PDF.

Sequence processing module -

Plot-file viewer module (and standalone PLT-viewer) v2.1b (with v8.0h:1) Increased number of corners The surface display in PLINFO.PLT could be wrong for planes with many corners (>100), the limit has been increased to 200. v2.1b (with v8.0h:1) Better memory handling of big plot-files

CATT-Edit v2.1b (with v8.0g:4) Two optional user templates for source file data

6

Addendum to Printed Manual for v8.0h:1 The text-files Template_src_1.txt and Template_src_2.txt in the CATTDATA folder can be used for custom source data that will inserted when the menu Template | Source | Template_src_1.txt or | Template_src_2.txt is selected.

CATT 3D-viewer v2.1c (with v8.0g:4) Markers back on v2.1b by mistake removed viewing of markers, now they are back again.

AutoCAD interface and DXF2GEO -

The FIReverb Suite (full version) PureVerb v4.0c (with v8.0h:1) DI and room dimension Gang options The Directivity Index and the Room dimensions have Gang options. For the room dimensions it changes all dimensions simultaneously with the same factor (drag any of the three sliders).

MultiVolver v4.0d (with v8.0h:1) Minor fixes Internal WAV-file library changes.

7

Dxf2Geo v1.0d Help

Copyright © CATT 2004-2007

Dxf2Geo is a small utility that converts AutoCAD® ASCII DXF™ files to CATT-Acoustic™ GEO-files. Although heavily associated with how AutoCAD works internally, DXF is also created and read by many other CAD programs since it is an open format so it can be useful to test other alternatives. The DXF surface entities handled are 3DFACE, 3D polyface mesh, LINE (when extruded to plane) and optionally closed polylines (POLYLINE, LWPOLYLINE and POLYLINE3D can be treated as if they were surfaces and will create planes). Further items drawn in the 3D window for reference are lines (such as pure LINEs) and surface colors. Solid-model DXF exports are not supported but have to first be converted to surfaces inside AutoCAD. So far, it is safest to export as DXF release 12 and to explode any blocks used.

1. First read a DXF-file considering these options: Treat POLYLINEs... Treat polylines as if they were planes. This can be a useful option to get away from 3DFACEs.

Unit used in file. If, after import, the unit is found to be wrong (see scale on top of z axis), select the proper unit and re-read the DXF.

Mark 3DFACE edges... Marks all 3DFACE edges (grey) or only those that have the visibility flag set (black).

Read DXF. Selects the DXF-file to read.

2. Then transform the geometry to make it suited for CATT-Acoustic use before saving as GEO: Rotate. Rotates the model the selected step clockwise (+) or anti-clockwise (-) around the axis indicated.

Translate. Translates the model in selected steps along the axis indicated. Center coord sys. Useful if the AutoCAD coordinate system origin is far outside the hall.

Reverse normals. A useful option if most normals seem wrong, or to visualize better.

Note: if you are planning to go back into AutoCAD, changing or correcting something, transform the geometry inside AutoCAD instead, or use this transform on the final export only. To aid in understanding the DXF data and see which surfaces exist, some viewing options are given: Entities. Indicates each surface and shows the DXF entity type.

POLY outlines. Normally, a POLYLINE treated as a surface is just shaded, this option also outlines its edges.

Both sides. Shades both sides of a surface (to better see where they are if the normals are inconsistent).

Warps. If warped planes are found they are outlined in red (warp > D/1000) and green (warp > D/10000) where D is the room diagonal. If no significant warps are found this checkbox is disabled.

All products mentioned are trademarks of their respective owners

Layers. Displays a dialog (see next page) where layers can be toggled on and off. If no layers are found this button is disabled.

Dxf2Geo Help

List of layers. RED indicates that the layer is on, BLUE that it is off. To toggle the state, double-click or press Enter (when a layer is selected). Selected layers are drawn with thick lines while the dialog is open. Only layers that are on will be exported.

Hides all layers (all on).

Shows all layers (all off).


Reverses normals for all surfaces in the selected layer. Closes the dialog export keeping the layers on/off and normal selections.

3. Finally, create CATT-Acoustic GEO-files and optionally source- and receiver-files (SRC.LOC and REC.LOC). If a re-exported DXF is to be converted the receiver- and source-files may already exist: File org.: Save all planes to the Master-file (Single), to ranges of 100 planes in each (id Ranges) or to one file per layer (Layers).

Include entites… Include, as comments, the entity (e.g. 3DFACE) that created a plane. Save non-face… Saves any pure LINEs found as markers in a separate GEO-file.

Material info. Unique ABS materials can be created from surface Colors or Layers and be given real values in GEO. Create … creates also generic receiver and source files. Save as GEO. Saves the currently viewed file to GEO. Note: the Transform options will be applied as shown in the 3D window and only layers that are on will be exported.

The master GEO-file name can be selected when saving, and will contain all corners and all ABS materials defined (mtrl_nnn if the Colors option is used, where nnn is the AutoCAD color number, layername spaces replaced with underscore if the Layers option is used, else a single material default. The remaining data depends on the File organization options selected: Single: both planes and markers (if requested) will be written to the single master-file. id Ranges: planes are written to auto-named pnnnn_mmmm.GEO files with 100 planes in each (nnnnmmmm is the range of plane ids in each file). If markers are requested, they are saved to a file pMARKERS.GEO and. The master then INCLUDEs the id range plane-files and the marker-file. Layers: planes are written to one file per layer layername.GEO. If markers are requested, they are saved to a file pMARKERS.GEO and. The master then INCLUDEs the layer-files and the marker-file. For each plane, the name is set to LINEnnnnn where nnnnn is the line in the DXF-file where the corresponding entity was found, and after each plane definition optionally the corresponding AutoCAD entity (e.g. 3DFACE) is given. At the moment this is mainly for debugging purposes and may be removed or made optional. All products mentioned are trademarks of their respective owners

2

Dxf2Geo Help


4. After a successful Save as GEO After a successful conversion and save to GEO, the remaining work can be anything from almost nothing to an amount of work that easily can make it slower to use a DXF-file than to make a model from scratch in GEO. It all depends on how well the model is made, how complex it is, and how suitable it is for geometrical acoustics modeling. It is also quite common that files made in AutoCAD are inaccurate (warped planes or gaps) since they have been modeled ”by eye” rather than by using GUI tools like ”snap” etc. It cannot be stressed enough: a model for geometrical acoustics must not be too detailed, but it must be accurate in itself to allow accurate predictions. The work after conversion to GEO will typically belong to one of two categories: 1. You have access to, and knowledge of, AutoCAD: In this case, when errors in the GEO are found, it is often best to go back into AutoCAD, make required changes, re-export and convert. If the Colors or Layers material options is used it will also save time when assigning materials in GEO. You may also model just half the room in AutoCAD and later use the GEO MIRROR directive. 2. You do not have access to AutoCAD but have a DXF-file only: In this case you are left to do all alterations and fixes in the GEO-files, a task that may be difficult since a DXF-conversion can only create a ”numbers-only” GEO-files. If the drawing has layers there is some help for alogical structure in GEO using the Layers file organization option. Problems It may well happen that some planes are detected as problematic. One way of handling that is to create a new GEO-file problems.GEO and cut/paste all problem planes into this file. When the majority of the planes have been read without immediate errors and can be displayed, problems.GEO can be INCLUDEd in the master-file and these planes can be handled one at a time by gradually moving a RETURN statement down through the list of problem planes. It may also happen that many planes must be reversed, removed or otherwise changed. If all, or most, planes seem reversed, either use the Reverse all normals option or change between FROMFRONT and FROMBACK in each file with planes. Note: It is very strongly recommended that a debug file is created before attempting any calculations what so ever after a DXF-file conversion. Missing data Even if a DXF-file converts well, a number of items are lacking simply because this information is not available in a DXF: Materials: actual values for ABS materials are lacking and any created materials have generic values only and. If the Colors or Layers material ootions could be used, it is sufficient to assign actual existing surface property names or give direct coefficients to each auto-created color- or layer-based material. If these options could not be used every single plane has to be assigned an appropriate material. Receivers: real data in the receiver file is lacking (if the option to create a receiver-file was chosen, it contains only a generic receiver position to enable viewing the model). Sources: real data in the source-file is lacking, likewise, if created, it contains only a generic source.


3

CATT WalkDecim v1.0c Help 1. Introduction WalkDecim is a small utility written to work together with CATT-Acoustic™ and 3D rendering software such as 3ds max®. The general procedure is: • • • •

•

to model a room in a 3D rendering program and export to CATT-Acoustic, or to export a model created in CATT-Acoustic to the 3D rendering program and graphicaly enhance it. create a camera path for a video walkthrough in the 3D rendering software. export the camera path to a text-file using a 3D rendering software script untility. import the path text-file to WalkDecim that then creates a CATT-Acoustic receiver-file with receivers along the path, as well as a complete Walkthrough Convolution Script (.WCS) for the CATT Walkthrough Convolver. after prediction and post-processing using these receivers, the Walkthrough Convolver can directly be used with the WCS script to create an auralized WAV-file that will be synchronized to a rendered video based on the indicated frame-rate used in the camera path export.

2. The exported camera path A camera path export utility must be created in the 3D rendering software (in 3ds max via MaxScript, other software have similar utilities) that exports camera and target positions to a text-file using the following TAB- or space-separated fields: n …

xc yc zc

xt yt xt

where: n xc yc zc xt yt zt

frame number (for reference, may start at any number) camera x-position camera y-position camera z-position target x-position target y-position target z-position

An example: 1 7.64889 -3.29267 2 7.64222 -3.20721 3 7.63514 -3.11644 4 7.62784 -3.02131 5 7.62033 -2.92252 … more frames

0.768435 0.789552 0.811958 0.835414 0.859747

7.51677 7.51306 7.51082 7.51014 7.51105

-1.091690 -0.928482 -0.769308 -0.614298 -0.463577

1.29888 1.33621 1.37213 1.40663 1.43965

3. How to use WalkDecim To read an exported camera path, first select the Unit used in the exported file and give the video Framerate, then click Read camera…. The camera path file selected (say path.txt) will then be read and the frame positions will be decimated to a number of receiver positions, based on the current selected maximum Move and Turn steps, and displayed in the 3D window. If the unit was set wrong, see scale on path drawing, just select the correct unit and read the file again. A new receiver position is created each time the camera has moved the specified Move distance or has turned the specified Turn angle (since the latest receiver was created). Typical good values are 1 m and 5° but if that creates > 100 receivers it is better to

CATT WalkDecim help


2

increase the Move distance than to increase the Turn value since too big turns will create jumps in the sound. CATT-Acoustic currently allows 00..99 receivers in one calculation, if too many receivers are created the last id is shown in red, else in green, and Save is disabled (if it is necessary to create more than 100 receivers please contact CATT, it is possible but requires some extra work). To see the new decimated path after changed parameters press Test, when satisfied give the Id for source used and the SIM-file folder (folder where created SIM-files will reside, typically the prediction output folder) and press Save receiverfile…. The first file created is the receiver-file path_rec.loc, (in the folder where the camera path was) where the relative and actual frame numbers are given as a comment. An example: RECEIVERS 0 7.649 -3.293 1 7.575 -2.276 2 7.524 -1.329 3 7.510 -0.709 4 7.517 -0.202 … more receivers

0.768 1.018 1.244 1.386 1.496

7.517 7.553 7.718 7.890 8.053

-1.092 0.344 1.140 1.467 1.729

1.299 1.604 1.726 1.740 1.740

; ; ; ; ;

1 11 19 24 28

1 11 19 24 28

The second file created is path.wcs for the Walkthrough Convolver (also created in the folder where the camera path was), a corresponding example: CATT-WCS VERSION = 0 SIMFOLDER = "c:\myroom\output" INCHANNELS = 2 INCHANNELSUFFIX = "_L" "_R" OUTCHANNELASSIGN = stereo MODE = time_absolute UNIT = frames 30 ;(fps) 1 "E_A0_00" 11 "E_A0_01" 19 "E_A0_02" 24 "E_A0_03" 28 "E_A0_04" 31 "E_A0_05" … more steps

Note that the MODE must be time_absolute and that UNIT must be frames (frames per second, fps). The frame rate must be the same as was used in the camera path export and in the actual video generation, this is necessary to synchronize the sound. WalkDecim sets these fields correctly.

4. Step by step A step by step procedure example how to how to create a video with binaural auralized walkthrough sound: 1

3D-rendering: Create a room model in the 3D rendering program and export to CATT-Acoustic. Or, if a CATT-Acoustic model already exists, it can often be exported as VRML, imported to the 3D rendering program and further enhanced graphically.

2

CATT-A: Perform some preliminary predictions to verify the model and its indented acoustic properties. Select the (stationary) sound source you want to use (say A0) to correspond to the sound source in the video.

3

3D-rendering: Design a camera path to be used for a walkthrough video with auralization.

CATT WalkDecim help


4

3D-rendering: Export the camera path using an appropriate utility, use the actual framerate of the video-to-be.

5

WalkDecim: Read the exported path file using WalkDecim and decide on a suitable path decimation. When satisfied enter the Framerate, the Id for source used (from step 2), the SIM-folder (the output folder from Step 2) and then Save reciver-file….

6

Copy the created receiver file (…_rec.loc) to the input folder used in step 2.

7

CATT-A: Select the Prediction module:

3

a) load the previous Prediction settings-file (PRD) for this model from step 2; b) in General settings select the receiver file created in step 5 and select as many receivers as are contained in the path (Which to use); c) select and edit Full detailed calculation and after setting appropriate calculation parameters (should have been found out in step 2) switch on (Save data for) Post-processing; d) Save and Run the prediction. This creates echogram ECH-files for each receiver to be used for binaural post-processing. 8

CATT-A: Open the Post-processing module: Select General settings suitable for the room case: as usual, IR length should be the same length as the longest reverberation time (if it differs a lot over the octave bands choose a length corresponding to the 1 kHz rev. time). Select binaural receiver (and if not already selected, HRTF and headphone equalization Prefs). Save the PST settings-file.

9

CATT-A: Open the Sequence processing module: a) add a REPEAT step that goes over all receivers using a variable RR; b) add a Post-processing step after REPEAT and select the PST settings-file from step 9 and one of the ECH-files created in step 8. Replace the receiver number in the ECH-file name with (if the selected file is e.g. …\E_A0_01.ECH change it to …\E_A0_.ECH). c) Save and Run the sequence. This creates binaural room impulse responses (SIM-files) for all receivers.

10

CATT-A: Reopen the Post-processing module (it was closed while doing sequence processing). Click Walkthrough convolution and Browse for the WCS-file created in step 5. Select an anechoic WAV-file and a resulting WAV-file name. Press OK.

11

After finished processing, the created WAV-file can be directly attached from start of the rendered video (using e.g. Adobe Premiere®) and should be perfectly synchronized.


CATT-WalkerTM v1.1f Help 1. Introduction CATT-WalkerTM creates real time auralization by continuously interpolating between B-format (W,X,Y,Z) impulse responses while convolving in real time creating a seamless simulation while walking around in the virtual room. The output is either binaural via a binaural down-mix of an ambisonic decode or B-format suitable for sub-sequent decoding. To be able to achieve low enough latency, CATT licenses a partitioned FIR convolution technique from Lake Technology. The calculations to use the real time walkthrough via CATT-WalkerTM is basically the same as using static auralization with a few exceptions: •

many receiver positions should be used, the more/denser the better. It is especially important around the source if a walkthrough close to the source is desired. In general, where the acoustics (i.e. the impulse responses) are expected to change fast with listener position or head direction, a high density is required while e.g. at the back of a long room it is less crucial. In the receiver-file a function recwalk() can be used to create positions in concentric circles around a source and for positions further back recloop() or recloop2() can be used. Below is an example of a good solution for a church:

41

42

43

44

45

46

47

48

32 29 40 49 33 15 141312 11 28 16 10 27 34 17 0203 09 A0 57 0001 08 26 35 18 19 07 06 25 2122 36 20 0405 39 65 37 24 38 23

50

51

52

53

54

55

56

58

59

60

61

62

63

64

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

31

30

Y

X

GLOBAL sx = 0 ;source x GLOBAL sy = -1.8 ;source y GLOBAL sz = 1.7 ;source z LOCAL rh = sz LOCAL da = 20 ;angle step for recwalk() LOCAL dr = 2 ;radius step for recwalk() LOCAL rfrac = 0.25 LOCAL dy = 3 ;y step for recloop() LOCAL dx = 3 ;x step for recloop() RECEIVERS 0 sx+rfrac*dr sy-rfrac*dr rh 1 sx+rfrac*dr sy+rfrac*dr rh 2 (-(sx+rfrac*dr)) sy-rfrac*dr rh 3 (-(sx+rfrac*dr)) sy+rfrac*dr rh recwalk(4,sx,sy,da,dr,-3.8,-10,3.8,3.7,rh) recloop(41,q,0,7,1, -2*dx,4.9+q*dy,rh) recloop(49,q,0,7,1, -1*dx,4.9+q*dy,rh) recloop(57,q,0,7,1, -0*dx,4.9+q*dy,rh) recloop(65,q,0,7,1, 1*dx,4.9+q*dy,rh) recloop(73,q,0,7,1, 2*dx,4.9+q*dy,rh) •

the head direction chosen in Prediction General settings does not matter when postprocessing for CATT-Walker since it is set as required during the special post-processing,

CATT-WalkerTM help


2

see below. The head direction can be chosen to give good binaural static auralizations or so that LEF values are according to standard (direction towards the source). •

a new file type is used: a CAG-file (CATT-Acoustic Geometry) that includes the actual geometry and all sources and receivers used for prediction. This file is always created in the output folder when Save data for post-processing has been selected. CATT-Walker uses this information to draw the real time auralization walkthrough 3D scene.

•

a special post-processing mode: Create CWI-file for CATT-Walker is used that does not create individual SIM- or PLT-files but instead a single big CWI-file (CATT-Walker Impulse responses) per source. This mode is enabled if Receiver model:B-format is selected in General settings. A CWI-file contains impulse response information for all predicted receiver positions for the source indicated in the file name. The IRs in an CWIfile can be displayed much like individual SIM-files using the Utilities|Information|CATT CWI-file menu. If details for specific receiver positions are interesting a normal postprocessing can be performed for those positions.

•

Since CATT-Walker is a separate program it does not read HRTFs and Headphone eq from CATT-Acoustic Preferences. Instead, for the binural ouput, a small program MakeWalkerBinDec.EXE has to be run to select the preferred filters that will be saved as a file BinDec.DAT in the CATTDATA folder, this file is then used by CATT-Walker. The initial BinDec.DAT is for 44100 Hz and uses the CATT1 HRTFs and Beyer DT 990 Pro headphone eq. Note: To change filters MakeWalkerBinDec has to be run before CATT-Walker is started but it needs to be run only once unless a new change in headphone or HRTFs is desired.

Below is an overview of the processing done and the files used by CATT-Walker: Anechoic WAV *.CWI

Prediction

E_ss_rr.ECH

Postprocessing for CATT-Walker

CATT-Acoustic

Walker.CAG

CATT-Acoustic

HRTFs Headphone eq

. . .

L CWI-file addition (delay, gain)

Walker_ss.CWI 1) Walker.CAG

MakeWalkerBinDec

CATTDATA

1) 2)

CWI 2)

Single source walkthrough Multiple source walkthrough via CWI-file addition

Real time R walkthrough auralization W CATT-Walker

X Y Z

BinDec.DAT Walker.CFG CATTDATA

External decoder for loudspeakers

. ..

CATT-WalkerTM help


2. Interface 2.1 Configuration When CATT-Walker is first started, a dialog is shown with license and version information:

Abort

aborts CATT-Walker and exits.

Help

displays this file

Start

starts the processing based on the settings chosen in Config.

Config shows a dialog that allows for selection of files, output and processing parameters. This dialog can stay open during processing and settings can be changed. However, all settings except Output Queue will force reloading of the filters.

3

CATT-WalkerTM help


4

Config: Audio group Output Format:

Binaural or B-format. B-format requires a 4-channel capable soundcard to be installed and selected and an ambisonic decoder to be used externally.

Output Device:

if more than one soundcard is installed this allows selection of which to use.

Block size:

the block size determines how many sample frames the system will process each iteration. The smaller the block size, the quicker the system will respond to interactive input. However, smaller block sizes also consume more CPU, and the audio driver may not be able to feed the blocks fast enough. 1024 is usually a reasonable compromise. The setting is PC hardware and IR length dependent, experiment and select a setting that does not give an interrupted sound and then click Save.

Output Queue:

this setting determines how many blocks will be queued by the audio driver. Generally a high value makes the playback more tolerant, but also increases latency. Try to keep it low if possible, say 3. The setting is PC hardware and IR length dependent, experiment and select a setting that does not give an interrupted sound and then click Save.

Approx Latency: based on the above settings, an approximate latency is estimated. The shorter the latency the faster the audio is updated during the walkthrough. 3D (Enable Z):

If the CWI-file contains the Z-component it is used for height information so that e.g. a source that is higher than a receiver it will perceived as higher up. If a particular PC hardware is on the limit to handle a case, the full 3D handling can be switched off (followed by Apply) to save some CPU.

Config: Scene group CAG:

the CAG-file to use (CATT-Acoustic Geometry). A CAG-file describes the room geometry in question and is created by CATT-A when saving data for postprocessing.

CWI:

the CWI-file to use (CATT-Walker Impulse responses). A CWI-file contains impulse response information for all predicted receiver positions for the source indicated in the file name. It is created when using the special post-processing option Create CWI-file for CATT-Walker.

WAV:

the anechoically recorded input WAV-file to use for the real time walkthrough convolution, if the file is stereo it will be mixed to mono before use. Note: the samplerate of the WAV must match the one used at post-processing.

Config: Save Saves the current settings to Walker.CFG in the CATTDATA folder. These settings are then automatically loaded the next time CATT-Walker is run. Config: Apply (labeled Start when Config is opened before processing) Applies the current settings for use when processing but without saving.

CATT-WalkerTM help


5

Start initiates loading and pre-processing of all IRs and the 3D geometry as indicated by a progress bar. When finished, a clean 3D window is shown and by clicking the mouse the walkthrough can start, by clicking the mouse again on the window you can go back to other applications or to the Config dialog. See next section for how to navigate and use other options. 2.2 Operation The walkthrough is controlled using the keyboard and the mouse together. The mouse is used to turn the head while the movement is controlled by pressing and holding the following keys or arrow keys: W or ⇑ S or ⇓ A or ⇐ D or ⇒ Q E +/Pg-Up / Dn

move forward move backward move left move right tilt right (rotation around view axis) tilt left increase / decrease walking speed (2 m/s is the speed at startup) raise / lower the sound level (master gain)

The movement is always at the height of the receivers and is restricted to be within the map of receivers as a whole. Some special keys are used e.g. to toggle on/off information in front of the 3D scene: Y H J G

C T F1 Q or ESC 1)

informative text a map of all receivers, current position and head direction, lines to all sources the very early part of the interpolated WXYZ IRs the wire-frame underlying the shaded graphics 1) Surface colors on/off textures on/off (for special purchased versions using textured graphics) opens the Config dialog if it was closed exit CATT-Walker

for special purchased versions (see 3.4), using textured graphics, the ”acoustic geometry” may typically be a simplified version and can be interesting to display for reference

3. News, limitations and possible future additions 3.1 Version history v1.0a First release version. v1.1a •

adds simulation of the height of the sound source (the B-format Z-component). This requires that CWI-files are created with CATT-Acoustic v8.0d:3 or higher. It also requires MakeWalkerBinDec.EXE v1.0b to be run again to create a new BinDec.DAT that holds info needed to decode height.

•

adds the option to simulate multiple PA system source where all sources have the same input sound. Similar to SIM-file addition, CWI-file addition can be performed using the

CATT-WalkerTM help


6

Utilities|CWI-file Addition menu. No matter how many CWI-files (sources) are added, the same amount of CATT-Walker CPU is required as for one source. Note that for CWIfiles to be added, they all have to be created from ECH-files saved at the same time using a multiple source prediction, this is to ensure that incompatible CWI-files are not mixed. The Multiple source addition module does not have to be used unless corresponding numerical measures are desired. Note that to resolve source interference at e.g. 1kHz a receiver spacing of half a wavelength or less at 1 kHz (17 cm) is required so don’t over interpret interference effects (or lack thereof) for big receiver spacings. Same principles holds for resolving fine directivity details, the walkthrough cannot give finder details than the receiver spacing allows for. •

Adds an option to select another soundcard than the system default.

•

Adds tilt controls (Q and E keys), only has effect on the sound if the Z-component is used.

•

Adds F1 to open the Config dialog if it was closed

•

Walking speed can be changed by the +/- keys (⇑⇓⇐ ⇒ and W S A D now use the same speed)

•

At first startup the Config dialog has a Start button that directly launches the 3D window and starts the processing, this button then changes to Apply and is used after loading a new CWI-file or after a change of some other settings.

v1.1b,c •

Minor fixes.

v1.1d •

Adds support for Walker Steer, see below.

v1.1e •

Changes to handle that from CATT-Acoustic v8.0g all data can be in a separate folder (called CATTDATA selected with File|Preferences), here it concerns BinDec.DAT and Walker.CFG.

v1.1f •

Optional B-format output for external loudspeaker decoding.

3.2 General limitations For positions close to and around the source it is important, as indicated in the Introduction, to use many receivers. However, for positions closer to the source than the inner ring of receivers the perceived sound source direction will be wrong (in the example when being inside the square formed by receivers 01,02,03 and 04). In practice it is of little importance since walking that close to a source is not very interesting.

32 33 14131211 15 16 1 4 17 0203 0 A0 0001 0 5 18 19 0 20 06 2122 36 0405 37

CATT-WalkerTM help


7

3.3 limitations and possible future additions The first versions concentrate on performance issues and lacks some implementation details that will be added gradually with updates depending on how the software is mostly applied and how well the many varying hardware combinations will perform. •

If a pillar or a wall is between receiver positions, CATT-Walker will interpolate and let you walk right through the wall. The intended application is not room mazes and similar, that are very difficult to predict and auralize well, but rather typical mainly open-plan rooms like auditoria, concert halls, churches, operas etc.

•

there is no overload indication, simplest is to move to a position closest to the source and use the Pg-Up/Dn keys to set a high level without overload (distortion) and then keep it like that.

•

actual source visuals are not used, the source is just indicated with a box and with a green arrow pointing in the forward direction and a blue arrow pointing upwards.

•

A Doppler effect is implemented but is disabled in the initial versions pending some further fine-tuning (it also consumes more CPU). Note that it anyway often can be better not to use Doppler for pure room acoustics walkthroughs since de-tuning of instrument tones can be very disturbing.

•

Windows multi-media drivers are used for the sound output, a future version may instead use ASIO drivers for lower latency.

•

multiple sources using different anechoic input can not be simulated. Future developments may be able to use further separate and independent sources simultaneously but will consume CPU in direct proportion to the number of sources.

•

direct ambisonic output for loudspeaker replay may be implemented in a future version.

3.4 Special versions Special purpose additions and versions of CATT-Walker will be possible but will require a specially purchased additional license. Due to the specialized nature of some of these additions it sometimes will have to be quoted as a dedicated project. Examples of available special versions are: •

external steering letting CATT-Walker act as an ”acoustics engine” in a virtual reality system where the graphics is rendered separately at higher detail. This is accomplished via the Walker Steer API/SDK that allows remote control of CATT-Walker. Details of how the remote control is implemented is up to the user (e.g. via TCP/IP if the graphics application runs on another computer or via a memory-mapped file if it runs on the same PC as Walker) but support is available on a case by case basis.

Examples of possible future special versions are: •

head-tracking, for normal use in front of a small screen head tracking is not crucial but for large VR screens it can be.

•

use of detailed textured graphics in CATT-Walker. This requires a detailed textured model created in 3ds max (but also other rendering software can be possible). The rendering

CATT-WalkerTM help


8

software will then via a plugin export to a special format used by CATT-Walker. The textured model can be fully detailed and does not have to use the simplifications of the geometrical acoustics model, it just has to use the same scale and coordinate system. If no detailed model exists a CATT-Acoustic model can be exported to VRML, imported into 3ds max and enhanced.


CATT Manual v8

Recommend Documents