Blind source separation using time-frequency distribution

In this paper, a new approach based on the signal´s time-frequency characterization, is described for blind source separation. Only two types of TFD algorithms are considered in signal separation applications, namely the short time Fourier transform STFT and Wigner-Ville distribution WVD. The main features of this approach apart from being suitable for nonstationary signals lie in the fact that it works satisfactorily even when the time-frequency signatures of the sources, overlap, or when the mixing system is convolutional and no longer linear. Illustrative examples are given to show that the proposed approach can successfully separate mixed signals when other approaches fail.