Speech quality enhancement using a two channel sparse adaptive filtering approach

A forward two channel blind source separation scheme can be used as a effective method for speech enhancement. In such a scheme, two mixed sound signals are used as inputs to estimate the original signals which created these mixtures. In an endeavour to enhance the speech quality, an attempt has been made in this paper to design a two channel blind source separation method based on a sparse learning strategy. The proposed approach takes advantage of the sparse nature of the acoustic path impulse responses of the mixing model and offers improved speech quality in comparison with traditional approach. Simulation study shows the improvement in speech quality in terms of perceptual evaluation of speech quality by the proposed sparse forward two channel blind source separation scheme.