Regularized NLMS Adaptive Algorithm for Noise Cancellation in Speech Signals

Abstract

Adaptive noise cancellation (ANC) is widely used to reduce noise from a noisy speech signal .However the Least Mean Square algorithm and its variants, such as the Least Mean Square Algorithm (LMS), normalized (N) - LMS algorithm and the constrained stability (CS)-LMS algorithms do not perform well in ANC since the desired speech signal has a bad effect on the convergence rate and steady state misadjustments of these algorithms. Thus, we propose a new adaptive algorithm that further relaxes the constrained in the CS- LMS algorithm .The new algorithm attempts to minimize the estimation error of the a posteriori error and the estimation is obtained using the concept of regularization .The analysis and simulation results show that the proposed new algorithm outperforms the LMS, NLMS and CS -LMS algorithm. Index Terms:Adaptive noise canceller, least-mean square (LMS) algorithm, Normalised LMS algorithm speech enhancement, stability constraint