COMPUTATIONAL MODELING OF AUDITORY STREAMING PHENOMENA

A computer model is described that uses simple physiological principle s that operate mainly at a peripheral level to account for perceptual coherence among successive pure tones of changing frequency. Using a s ingle set of parameter values, the model is able to reproduce a number of fundamental auditory streaming phenomena. These include the build- up of auditory stream segregation over time, and the temporal coherenc e and fission boundaries of human listeners. Whereas these streaming p henomena are generally accounted for in terms of a high-level auditory scene-analysis process, the success of the model in reproducing exper imental data obtained from humans justifies the potential value of a l ow-level analysis for explaining auditory grouping phenomena, and sugg ests that some auditory grouping may be the product of low-level audit ory processing.