Neural-Scaled Entropy
The goal of this project is to model
differences in speech perception associated with different digital signal
processing techniques for hearing aids (e.g., frequency
lowering, wide
dynamic range compression) so that we can better understand the
complex interactions between the hearing loss and the acoustic signal that
determine whether an individual will be more or less successful when
communicating. The starting
framework is the Cochlea-Scaled Entropy model (e.g., Stilp,
Kiefte, Alexander, and Kluender 2010), which in turn, was inspired by
theories of perception that embodied Shannon Information (Barlow, Attneave, etc.).
A similar information-theoretic approach is applied here to neural
representations of speech, which model adaptation as well as broadened auditory
filter tuning associated with hearing loss and increased presentation
level.
|
