Speech processing by a model of the auditory periphery

A composite, computational model of the auditory periphery, from the tympanic membrane to the synaptic junctions between hair cells and Eighth Nerve fibers, is discussed. The model includes linear and nonlinear processing steps known to occur in the periphery. Synthesized acoustic signals are converted into a population of responses that can be directly compared with neural data. Although components of the model have been previously tested only with simple acoustic signals, the model correctly predicts many properties of neural responses to vowels. In particular, the simulated rate responses saturate at moderate stimulus levels, while remaining strongly synchronized to vowel formant peaks.