Blind separation of noisy convolutive sources

In this paper, we present a new blind source separation method for noisy linear convolutive signal mixtures of independent components. The effectiveness of the proposed method is tested on synthetic data. Our technique has various advantages. First, it is based on the expectation-maximization (EM) algorithm for the separation of the components. Second, the proposed technique works in the spectral domain where, thanks to two simple approximations, the likelihood assumes a simple form which is easy to handle (low dimensional sufficient statistics) and to maximize (via the EM algorithm). Using this technique, we have obtained a good preliminary results being able to blindly separate noisy mixtures with two components and four different versions of mixing matrix.