Real-time microphone array processing for sound source separation and localization
Longji Sun, Qi Cheng · 2013
In this paper, the problem of sound source separation and localization is studied using a microphone array. A pure delay mixture model which is typical in outdoor environments is adopted. Our proposed approach utilizes the subspace method to estimate the directions of arrival (DOAs) of the sources from the collected mixtures. Since sound signals are generally considered broadband, the DOA estimates for a source at different frequencies are used to approximate the probability density function of the DOA. The maximum likelihood criterion is used to determine the final DOA estimate for the source. Using the estimated DOAs, the corresponding mixing and demixing matrices in the frequency domain are computed, and the source signals are recovered using the inverse short time Fourier transform (STFT). Our algorithm inherits the robustness to noise of the subspace method and also supports real-time implementation. Comprehensive simulations and experiments have been conducted to examine various aspects of the algorithm.