Automatic tap-length adjustment of adaptive filter for feedback attenuation in hearing aids

Abstract

In this paper, a tap-length-adaptation algorithm is proposed for adaptive feedback cancellation (AFC) in the adaptive filter used in digital hearing aids. The existing state-of-the-art feedback cancellers use a fixed-length adaptive filter, which might over-model or under-model the feedback path. To overcome the issue of selecting a fixed arbitrary tap-length value, a unique automatic tap-length-variation algorithm is proposed for the adaptive filter. A key tap-length-update parameter, which ensures a balance between the deviation between steady-state and optimal values of the tap length and influences the convergence speed of the tap-length adaptation, has also been made adaptive for efficient feedback path estimation. A detailed mathematical performance-analysis of the proposed algorithm at steady state is performed. Computer simulations verify that the derived expressions predict the proposed algorithm’s improved performance over that of the existing algorithms.