Unbiased adaptive feedback cancellation in hearing aids

Shusina, Ngwa Abinwi (2003) Unbiased adaptive feedback cancellation in hearing aids. University of Southampton, Doctoral Thesis.

Abstract

This thesis is concerned with feedback identification and cancellation in digital hearing aids where current methods such as the direct method produce biased feedback path estimates with limited feedback cancellation particularly for high gain hearing aids.  Artificially introduced delays work well for broadband signals but are futile for highly correlated signals.  Furthermore, large artificial delays introduce pre-echo and comb filter effects which reduce the audio quality.

In this thesis, unbiased closed-loop identification strategies for feedback cancellation are suggested, studied and developed in the context of hearing aids.  These methods have the advantage of minor dependence on the speech signal type and delays in the feed-forward path. Further, good modelling accuracy, and feedback path cancellation is obtained particularly at high frequencies where instability typically occurs for high gains.  The unbiased methods utilize an identification signal usually ‘white noise’, which is normally set below the hearing threshold of a hearing impaired individual.

It is shown in the thesis using realistic simulations of the hearing aid that the projection method which uses non-causality to identify the feedback path in two stages, manifests superior performance in contrast with the direct method specially during howling.  The bin-normalized frequency-domain LMS is consolidated with band-pass filtering to improve convergence speeds and mitigate disturbance signal problems which exacerbate mis-adjustment noise.

This record has no associated files available for download.

Contributors

Download statistics