Individuals with normal hearing are adept at understanding speech in the presence of noise, such as other speakers or environmental sounds. In contrast, individuals with hearing loss struggle to understand speech in the same adverse conditions. Neural processing in the inferior colliculus (IC) of the brainstem appears to contribute to the ability to separate simultaneous competing sounds. A computational model developed in the Sinex lab reproduces the responses of IC neurons to complex sound

more »

... tures. It seems likely that the model can be applied to improve the processing of speech in noise. The computational model's effectiveness at improving the processing of speech in noise is evaluated through a perceptual experiment which uses the model to process sentences that are then presented to listeners. The experiment's data are analyzed to evaluate the pattern of errors. The analysis shows that low frequency speech features are being accurately transmitted by the model while high frequency speech features are not. This pattern suggests ways in which the computational model may be improved. Possible technological and clinical applications of the computational model for individuals with hearing loss will also be discussed.