Effect of a Signal Decorrelation on Adaptive Filtering Algorithms for Acoustic Echo Cancellation

Acoustic echo cancellation (AEC) is a well-known issue in voice communication systems; it is used to reduce undesired echoes. Adaptive filtering is generally employed for AEC. However, the performances of adaptive filtering algorithms are limited for a correlated input signal. This paper presents the effect of an input signal decorrelation on the performances of adaptive filtering algorithms for AEC application. Therefore, normalized least mean squares (NLMS) algorithm is used, performed by: USASI noise and speech signals. Simulation results show an improvement in term of convergence speed of NLMS algorithm for the decorrelated signal compared to the original signal.