Multiple Sound Source Separation by Jointing Single Source Zone Detection and Linearly Constrained Minimum Variance

In this work, a multiple speech source separation method is proposed based on single source zone detection and linearly constrained minimum variance (LCMV) beamformer. First, the "single-source" zones where only one source is dominant over the others are detected according to the characteristic of B-format microphone. Second, the number of active sources and their corresponding directions of arrival (DOAs) are estimated in these "single-source" zones. Finally, LCMV beamformer is adopted for recovering the sources separately by using the DOA estimates. The objective evaluation shows that the proposed method achieves better separation quality compared to some existing BSS (blind source separation) approaches.