A Novel Double Talk Echo Canceller Algorithm Using Multi Delay Filter

Abstract

Echo is a major problem in voice communication system. Several methods have been proposed to overcome this problem. Recently, a fast converging algorithm called Block Robust Proportionate Normalized Least Mean Squares (BRPNLMS++) against double-talk has been proposed. This paper presents a novel algorithm called (MDF-BRPNLMS++) based on the BRPNLMS++and the Multi-Delay Filter (MDF) algorithms. The proposed method allows reducing considerably the execution time, due to the use of the MDF, and reaching faster convergence by estimating the update step-size at each block iteration. In addition, the proposed method does not require a Double- Talk Detector (DTD) which contributes in reducing the required computational power. Simulation results have been performed using Matlab and prove a good agreement with the theoretical results.