Underdetermined blind separation of non-sparse sources using spatial time-frequency distributions

Abstract

Recently, a number of underdetermined blind source separation (UBSS) approaches have been proposed to separate n source signals from m (m<n) mixtures. The majority of these methods depend on the assumption that the source signals are sparse in either time domain or other transformation domains. While the sparsity assumption is reasonable for some applications, such as the separation of speech signals, it is restrictive for many other applications. In this paper, we propose a new time-frequency UBSS algorithm without resort to any sparsity conditions. We show that under some mild constraints, our algorithm can separate up to 2m−1 sources from m (m⩾3) instantaneous linear mixtures. Theoretical analysis and simulation results show the effectiveness of the proposed algorithm.