16th International Workshop on Acoustic Signal Enhancement (IWAENC), Tokyo, Japan, 17 - 20 September 2018, pp.81-85
Acoustic source separation refers to the extraction of individual source signals from microphone array recordings of multiple sources made in multipath environments such as rooms. The most straightforward approach to acoustic source separation involves spatial filtering via beamforming. While beamforming works well for a few sources and under low reverberation, its performance diminishes for a high number of sources and/or high reverberation. An informed acoustic source separation method based on the application of orthogonal matching pursuit (OMP) on complex-valued steered response (SR) maps is presented in this paper. It is shown that the proposed method is robust to a high level of reverberation and affords better source separation than maximally directive steered beamforming in terms of signal-to-interference and signal-to-distortion ratios.