CATT-Acoustic v9.1

Modeling Module

The hall geometry is described using a structured hierarchic text format created manually, via AutoCAD or other CAD plugins:

INCLUDE statements enables a hierarchical structure with hall parts defined in separate files
expressions/math functions (e.g. sin(), cos())
local and global symbolic numerical constants (e.g. GLOBAL ceil_height = stage_height + 10 )
symbolic string constants (e.g. LOCAL wup = "wall of unknown plaster") e.g. shortcut names.
interactive input of constants with limit check (e.g. GETLOCAL h = 8 INFO = "Ceiling height in meter" MIN = 5 MAX = 18)
tracing and break statements (SAY, RETURN, BREAK, DEBUG, DEBUGON, DEBUGOFF)
if-then-else statements for geometry variants
IF balcony THEN
... balcony planes
ELSE (optional)
... other planes
ENDIF
comments
tools for surface corner locking and object rotation, translation and copying etc. (e.g. lock(), cut(), x(), y(), z(), OBJECT, ROTATE, TRANSLATE, COPY)
mirroring of symmetrical (or nearly symmetrical) models (MIRROR)
optional hiding of invisible edges
one-step generation of circular coordinates and similar using the loop() function ( e.g. loop(500,a,0,315,45,2*cos(a),2*sin(a),3) )
plane loops to create e.g. circular shapes with one command
double-sided planes with one command
one-step generation of receiver coordinates using the recloop() functions (possible OVERRIDE of loop-created positions)
enables very geometrically accurate, consistent and maintainable models
symbolic surface property names allowing for high-level contsructs like wall_abs = plaster making it possible to e.g. change all wall absorption by changing one definition (wall_abs = absorber)
special surface property definition for audience mapping plane
frequency dependent scattering via 2D Lambert or a novel 1D Lambert (for walls with battens and 1D diffusers)
frequency dependent semi-transparent surfaces (e.g. hanging curtains)
export of geometry to AutoCAD interface, DXF, VRML, CATT Marker 3D object format (for GEO objects and source visuals), CLF cabinet format (useful for CLF authors).
import of geometry via AutoCAD interface or DXF (DXF2GEO converter available also in demo)
merging of planes e.g. after an AutoCAD interface export or DXF with 3DFACEs
import of geometry via SketchUp plugins (3rd party: Sk2Geo by Euphonia, France, SU²CATT by K5 GmbH, Germany.)

simple shoebox GEO-files/graphics

an AutoCAD interface written in AutoLISP is included that creates files in the CATT format directly (models created in the CATT format can also be exported to AutoCAD via the interface using an intermediate file format)
Export to the CATT 3D-viewer based on OpenGL including source directivity -3, -6 and -9 dB contours. Interactive source aim and positioning, surface data and selection.
A tool for designing ceiling loudspeaker arrays for voice evacuation systems

ceiling loudspeaker array tool

Optional non-acoustical element "markers". Markers can be in the forms of lines, polygons, circles, discs, rulers, 3D text and loaded objects (e.g. chairs, etc.).

simple marker graphics

General utilities requiring no time-consuming calculation:

	source directivity contours
	reflection coverage

VRML 2.0 export. The Virtual Reality Modeling Language enables interactive 3D models - with auralized sound (click a 3D button and the binaural sound associated with the view is played). VRML-files (.wrl) can be linked to web pages by using a VRML Browser plug-in or add-on to WWW browsers. Note: VRML has since some time been replaced by the X3D format but one version of X3D is essentially VRML (minor header differences), for list of suitable browsers see http://www.web3d.org

	sample VRML scene, as exported, screen shots, *no VRML-plugin required*
	sample VRML scene, as exported, VRML-plugin required

VRML can also be imported into many 3D graphics software such as 3D Studio.

Interactive RT estimate calculation:

Included is a very powerful and easy to use spreadsheet-like function where Sabine and Eyring RT estimates can be done while changing absorption. In addition a global ray-tracing can be performed that contiously updates the RT-curve. For some halls (due to a "mixing" shape or suitable diffusion and absorption distribution) the Sabine and Eyring RT estimates come very close to actual T-30 values estimated from line regression on the decay curves. For such halls, what if? RT calculation games can quite safely be based on classical RT estimates. In addition to Undo and Revert options, the surface list can be sorted according to various criteria to aid in understanding which surfaces control the RT most:

absorption (125&250, 500&1k, 2k&4k or 125-4k Hz)

absorption Area (125&250, 500&1k, 2k&4k or 125-4k Hz)

plane (Id, Name or Area)

absorption use statistics (how much of each material is used etc.).

Sabine, Eyring

mean absoption coefficient

mean free path

volume and surface area

Schröder frequency

geometrical acoustics limit estimate

utilizes multi-core CPUs (a complex model will generally be twice as fast on a dual-core CPU and four times faster on a quad-core)

Sample RT-calculation dialogs

1995 prediction round-robin results:

methods employed in a previous version of CATT-Acoustic (v5) were very successful in a round-robin with 16 participants (Vorländer, Proc. 15th ICA Trondheim 1995)

CATT-Acoustic was one of only three programs that were judged to give reliable and useful results. Of these three 5 of the 8 predicted measures were best evaluated by CATT-Acoustic

The RTC utilized by v7/v8 gave overall better results being of a more general nature

The all-new CATT TUCT™ used by v9 adds to the capablities and replaces the RTC see overview.

CATT-Acoustic 1995 round-robin model

Now more than 25 years later this has to be discussed in a deeper way. At the time I was happy about this type of actually
quite limited verification (one room, but there had previously been several internal measurement comparisons) when users
were very uncertain about what GA software could do, and since so many had bad results at the time that worry was real.
The actual text in the paper was:

”Although the accuracy rating is somewhat arbitrary and should not be over-interpreted, it seems to be reasonable ”…
“there are only three programs that are unquestionably reliable in the prediction of room acoustical parameters; F, 8 and 1”.

Unfortunatelly this has many times been totally over-interpreted and assumed to also hold for all other rooms even very non-diffuse,
coupled and even open rooms such as Roman and Greek arena. The round-robin 1 room was only slightly non-diffuse (T30 slightly
longer than Sabine) and it seems totally forgotten (or for new software users 25 years later not knowing/checking it) that the test
was also only for 1 kHz at a time when one of the three software only had one scattering coefficient (applicable around 500-1k Hz).
Worse, the quote above has even in software PR been "expanded" and claimed to hold for all three round-robin.

In my view the only really meaningful test is for software users to compare their own predictions to measurements and not just once
but continously as new persons come in and start to model. In 2008 there was an official user round-robin (8 users of the same
sofftware) and the results varied enormously.