Design of a class of zero attraction based sparse adaptive feedback cancellers for assistive listening devices

Abstract

Acoustic feedback is a frequently encountered problem in assistive listening devices (ALDs). Feedback paths in ALDs are typically sparse in nature and sparsity aware adaptive feedback cancellers can improve perceived audio quality under such scenarios. In an endeavour to improve the feedback canceller performance, a decorrelated polynomial zero attraction (DPZA) normalized least mean square (NLMS) feedback canceller is proposed in this paper. DPZA-NLMS algorithm is seen to have higher computational complexity. Hence, in an attempt to reduce computational complexity, a decorrelated -NLMS (D--NLMS) and a decorrelated Versoria zero attraction NLMS (DVZA-NLMS) based feedback canceller are also proposed. Feedback canceller performance in terms of convergence and tracking performance as well as speech/audio quality and speech intelligibility is compared. In addition, computational complexity and memory requirements of the algorithms are also compared thus providing a hearing aid designer with better trade off choices between computational requirements and feedback cancellation performance.