DSP algorithms for digital hearing instruments.

A new digital filter bank design and a new compression algorithm that can improve the performance of hearing instruments located completely in the ear canal (CIC) are developed in the thesis. In order to assess state-of-the-art hearing instruments employing advanced signal processing techniques the DynamEQ-II analog hearing instrument developed by the Gennum Corporation was studied extensively. A sophisticated SIMULINK model, involving the use of audio files, was developed to evaluate the performance characteristics of the strategies and algorithms used in the DynamEQ-II. The RangeEar algorithm employed in the DigiFocus hearing instrument from the Oticon Company was also studied using SIMULINK in a similar manner. Two recommended improvements for a new hearing instrument are presented. The first improvement involves the use of an eight-band digital filter bank based on an interpolated finite impulse response (IFIR) prototype filter that has been optimized using delay elements to give a maximally flat overall magnitude response. The resulting group delay is a constant and less than the value where self-hearing and lip reading problems occur. The second improvement uses a new compression algorithm based on a model of the human auditory system. The new algorithm replaces the existing constant homomorphic multiplication algorithms with an acoustic signal intensity weighted multiplication. The resulting nonlinear compression ratio expands low level signals and compresses high level signals in such a manner so as to improve noise immunity and increase the intelligibility of the sound. The MIT hearing loss simulator was employed to evaluate the effectiveness of the new proposed filter bank and compression algorithm by analysis of and listening to actual test audio files.Dept. of Electrical and Computer Engineering. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2001 .O53. Source: Masters Abstracts International, Volume: 41-04, page: 1157. Adviser: W. C. Miller. Thesis (M.A.Sc.)--University of Windsor (Canada), 2001