Speech Analysis and Cognition Using Category-Dependent Features in a Model of the Central Auditory System