ISCA Archive Interspeech 2022
ISCA Archive Interspeech 2022

Multi-source wideband DOA estimation method by frequency focusing and error weighting

Jing Zhou, Changchun Bao

In this paper, a new multi-source wideband direction of arrival (MSW-DOA) estimation method is proposed for the signals with non-uniform distribution using the sub-array of uniform linear array. Different from conventional methods, based on the free far-field model, the proposed method mainly makes two contributions. One is that the sub-array decomposition is adopted to improve the accuracy of MSW-DOA estimation by minimizing the weighted error, and the other one is that the frequency focusing procedure is optimized according to the presence probability of sound sources for reducing the influence of the sub-bands with low signal to noise ratio (SNR). Simulation results show that the proposed method can effectively improve the performance of wideband DOA estimation in the case of multiple sound sources.