Development of System for Speech Enhancement using Combinational Adaptive LMS on Reconfigurable Platform

Abstract

Low cost and robustness makes the least mean square (LMS) algorithm most popular. Improved adaptive filter performance can be obtained by combinational approach. In this paper, for speech enhancement, we used one of such approaches known as ‘filter combination’. As our simulation results show, Fixed-Sign LMS algorithm overcome other algorithms in terms of SNR and mean square error.

Keywords: Adaptive filter, speech enhancement, windows, convex combination