Abstract
Cognitive modeling and processing for speech recognition - ears and beyond
Fred Juang
Georgia Institute of Technology
In this talk we discuss our recent work in modeling the auditory system from the ear to the primary auditory cortex for signal processing applications such as automatic speech recognition. We address the issue of redundancy in cognitive representations as stipulated by many researchers in the field. We present our findings which suggest an unconventional category or class dependent feature selection process that is shown to lead to robust speech recognition performances.
No video available