[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: CASA problems and solutions

Dear participants of the recent CASA discussion,

since I've finally managed two weeks ago to put an accompanying demonstration page for my "Identification and segregation of multiple auditory objects"-model (developed 1990-1994 as part of my doctoral thesis, supervised by E. Terhardt) on the web, some of you might be interested to visit


The ingredients for this approach to CASA are:

1.      van Noorden's sequential grouping observations
2.      Rasch's important reports on the role of onset asynchrony
3.      Some of Al Bregman's Gestalt motivated rules (summarized in his pamphlet)
4.      The continuity effect
5.      Terhardt's "Virtual Pitch" procedure to account for the "Seebeck problem"
6.      Some research of my own to investigate boundaries for mistuned, asynchronous  harmonics in a sequence of complex harmonic sounds
7.      And a lot of struggling with C-programming with double linked lists to form auditory objects and relations between them

The model disregards binaural as well as phase information. Also, no information about timbre is used for segregation.

Results: This approach seems to work  with fair results for "Helmholtzian" sounds and simple two part music, but complex signals with reverberation and overlap of onset and offset of sounds will give poor results.


Uwe Baumann

Dr.-Ing. Uwe Baumann
Klinikum Großhadern/HNO/Audiologie
D-81377 München, Marchioninistr. 15
Tel.: +49/89 7095 3878 FAX: +49/89 7095 6869