A Variable Step Size Improved Multiband-Structured Subband Adaptive Feedback Cancellation Scheme for Hearing Aids

Acoustic feedback degrades the sound quality and limits the maximum stable gain of modern open fitting behind the ear hearing aids. The standard adaptive feedback cancellation approach results in a biased estimation of the feedback path. Despite having several bias reduction methods, the convergence behaviour of the feedback canceller is deteriorated due to the highly correlated input signals such as speech and music. This paper proposes an improved delayless multiband-structured subband adaptive feedback cancellation scheme, which can be considered as a generalized form of the normalized least mean square algorithm, the affine projection algorithm and the multiband-structured subband algorithm. Both the prediction error method and frequency shifting are incorporated in the proposed scheme to reduce the bias. The improved proportionate technique along with a non-parametric variable step size technique is further developed to provide faster convergence and tracking rates. Simulation demonstrates improved transient and steady-state behaviour, improved tracking rate and enhanced speech quality.