Modified Adaptive Line Enhancer in Variable Noise Environments using Set-Membership Adaptive Algorithm

Abstract

In most speech processing systems, speech signals degrade significantly in noisy environments. Therefore, noise cancellation methods are necessary to eliminate noise from the speech signal. This paper presents a development technique to improve the performance of a noise cancellation system called Adaptive Line Enhancer (ALE) in variable noise environments. The method is based on using a modified and simplified version of the set-membership Affine Projection algorithm (AP) adaptive filter. The modification introduced here is to make the adaptive algorithm, work with a dynamically variable step-size, as a substitute for the commonly used Least Mean Square (LMS) algorithm in ALE system. The performance of the developed ALE in variable noise is illustrated here, and the results are compared to similar ALE systems based on the classical adaptive algorithms, such as the Least-Mean-Square (LMS), the Normalized LMS (NLMS) and the Affine Projection (AP) algorithms. The developed adaptive line enhancer using the simplified affine projection algorithm showed a better performance than those based on LMS, NLMS and AP.