Residual Cross-talk and Noise Suppression for Convolutive Blind Source Separation

Introduction Blind source separation (BSS) refers to the problem of recovering signals from several observed linear mixtures (e.g., [1]). In this paper we deal with the convolutive mixing case as encountered, e.g., in acoustic environments, and aim at finding a corresponding demixing system, whose output signals yq(n), q = 1, . . . , P are described by yq(n) = ∑P p=1 ∑L−1 κ=0 wpq,κxp(n−κ), and where wpq,κ, κ = 0, . . . , L−1 denote the current weights of the MIMO filter taps from the p-th sensor channel xp(n) to the qth output channel (Fig. 1). We assume that the number of active source signals Q is less or equal to the number of microphones P . BSS algorithms are solely based on the assumption of mutual statistical independence of the different source signals. The separation is achieved by forcing the output signals yq to be mutually statistically decoupled up to joint moments of a certain order. In