Spatial aliasing suppression for pooled angular spectrum using a widely spaced microphone array
Zhiyong Xu, Zhao Zhao · 2016
A time-frequency pooled angular spectrum capable of suppressing spatial aliasing effectively is studied for a widely spaced microphone array to estimate multi-source directions of arrival (DOA) in a reverberant environment based on diffuse noise model and time-frequency sparsity of acoustic signals. By using constant false-alarm rate (CFAR) detection technique, only the high-valued elements very likely corresponding to the active sources are temporally pooled. As a result, the difference between the true DOA peaks and the false ones are enlarged significantly, which is more important for multi-source detection and localization. Simulation results demonstrate that the proposed method obviously outperforms the existing counterparts in spatial aliasing suppression for widely spaced arrays.