Evaluation of Performance of Genetic Algorithm for Speech Signals Separation

Abstract

Blind source separation is an important issue for signals processing. In this paper, a blind source separation based on continuous and binary genetic algorithm is proposed. The proposed method includes several main steps of preprocessing which are centering, whitening and orthogonalization. The separate matrix is updated by high order statistics of kurtosis. Most of papers have been focused on three sources. But, in this paper the blind source separation for more than three sources is investigated. It is shown that continuous genetic algorithm considerably would be better to get result than binary genetic algorithm for solving the blind source separation problem for different number of sources with high accuracy, fast convergence performance and suitable SNR. In result enhanced separation of mixed signals plus noise or interference has obtained.