A New Approach for Blind Source Separation of Convolutive Sources - Wavelet Based Separation Using Shrinkage Function

ICA and its variations are used extensively in BSS. Most of the algorithms that are used to separate speech or music signals utilize ICA in the time frequency domain. Here ICA is applied in the wavelet domain. Separation of signals is achieved by applying the ICA algorithm and shrinkage functions to the wavelet coefficients of the original mixtures. ICA alone can achieve reasonably good separation of artificially convolved sources; however, poor separation quality is experienced for real world convolutive mixtures. This work presents a novel post processing technique to deal with the cross talk problem. The post processor is applied to the signals separated by the ICA network. A super Gaussian form of the PDF is assumed for the dominant source components. Closed form solutions of the parameters of the PDF are obtained by the MOM. The PDF of the cross talk components is assumed to be of a GMM, and the EM method is applied to determine the parameters of the Gaussian mixtures. The algorithm is applied to a real world mixture of music and speech signals. The results show a significant reduction in the cross talk.