gms | German Medical Science

Auditory perception models attempt to predict psychoacoustical data from a large variety of experimental conditions, typically inspired by insights from physiological coding principles along the auditory pathway as well as concepts of cognitive processing and decision making in the back end of the chain. The first part of the presentation will provide examples of computational model frameworks that have focused on accounting for resolution limits of auditory processing and perception as well as correlates of auditory stream segregation. The second part will expand the modeling approach towards predicting speech intelligibility in basic and advanced (adverse) acoustic conditions. Both approaches contain stages of feature selectivity and adaptation in the auditory processing as well as a decision metric based on either the signal-to-noise power ratio or a correlation measure. The proposed models are able to account for the effects of background noise, reverberation, nonlinear distortions and noise reduction processing on signal detection and speech intelligibility. However, the models have challenges to account for the consequences of individual hearing loss on the respective outcome measures. The goal of current research is to disentangle the consequences of different types of hearing deficits and to understand how behavioral goals define or re-tune auditory features at the cortical level.