Affine projection algorithm for acoustic feedback cancellation using prediction error method in hearing aids

Prediction error method (PEM) is popularly applied to acoustic feedback cancellation (AFC) in hearing aids. Commonly, this method uses normalized least mean square (NLMS) adaptive filter to estimate the coefficients of the real feedback path. A disadvantage of NLMS algorithm is to provide a slow con...

Full description

Bibliographic Details
Main Authors: Tran, Linh T.T., Dam, Hai Huyen Heidi, Nordholm, Sven
Format: Conference Paper
Published: 2016
Online Access:http://hdl.handle.net/20.500.11937/50643
Description
Summary:Prediction error method (PEM) is popularly applied to acoustic feedback cancellation (AFC) in hearing aids. Commonly, this method uses normalized least mean square (NLMS) adaptive filter to estimate the coefficients of the real feedback path. A disadvantage of NLMS algorithm is to provide a slow convergence rate when coloured incoming signals are used. To address this problem, we propose a simple but effective way to apply an affine projection algorithm (APA) to acoustic feedback cancellation using PEM. Performance of the proposed method is evaluated for speech incoming signal in both cases of using/not using a probe noise. Simulation results show that the proposed method outperforms the PEM using NLMS in both terms of misalignment and added stable gain.