Multiple Acoustic Source Localization in Microphone Array Networks

The problem of multiple acoustic source localization using observations from a microphone array network is investigated in this article. Multiple signals are assumed to be window-disjoint-orthogonal (WDO) on the time-frequency (TF) domain and time delay arrival (TDOA) measurements extracted at each TF bin. A Bayesian model then proposed jointly assign different sources estimate locations. Considering that WDO assumption usually violated under reverberant noisy environments, we construct relational by coding distance information between distributed arrays such adjacent have higher probabilities observing same source, which able mitigate miss detection issues adverse environments. Laplace approximate variational inference method introduced hidden variables model. Both simulations real data experiments performed. results show our achieve better accuracy than existing methods.