A unifying framework for acoustic localization

Recent advances in acoustic localization have combined the advantages of the traditional methods of beamforming and time-delay estimation, leading to techniques that are both accurate and fast. We present a unifying framework that reveals the relationships between beamforming, time-delay estimation, Bayesian formulation, hemisphere sampling, and accumulated correlation. We then experimentally compare the algorithms, on both compact and distributed microphone arrays, showing that the recent technique of accumulated correlation, although much less computationally expensive, exhibits performance comparable to that of beamforming.