Modeling Outcomes with Frequency Lowering using Neural-Scaled Entropy

 

Varsha Hariram

 

Committee:

Joshua Alexander, Ph.D. (Chair)

Michael Heinz, Ph.D.

Keith Kluender, Ph.D.

 

Signal processing schemes such as nonlinear frequency compression (NFC) recode speech information by moving high-frequency information to lower frequency regions.  Perceptual studies have shown that depending on the dominant speech sound, where compression occurs and the amount of compression can have a significant effect on perception.  Very little is understood about how frequency-lowered information is coded by the auditory periphery or even what constitutes “information.”  We have developed a measure that is sensitive to information in the altered speech signal in an attempt to predict optimal frequency lowering settings for individual hearing losses.  The Neural-Scaled Entropy (NSE) model examines the effects of frequency-lowered speech at the level of the inner hair cell synapse of the auditory nerve model [Zilany et al. 2009, J. Acoust. Soc. Am., 126, 2390-2412].  NSE quantifies the information available in speech by the degree to which the pattern of neural firing across frequency changes relative to its past history (entropy).  Nonsense syllables with different NFC parameters were processed through a variety simulated hearing losses using the auditory nerve model.  Results are compared to perceptual data across the NFC parameters as well as across different vowel-defining parameters, consonant features, and talker gender.