Localizing an unknown time-varying number of speakers: a Bayesian random finite set approach

Using time-difference-of-arrival (TDOA) measurements to perform speaker localization has received much interest recently. Motivated by the significant progress in TDOA single speaker localization, this paper presents a TDOA multi-speaker location tracking algorithm based on Bayesian particle filtering. The development is based on the random finite set framework, which provides an effective treatment to the problem of an unknown time-varying number of active speakers. The proposed method can be viewed as a generalization of the existing single-speaker particle filter. Using a simulated reverberant room, we demonstrate the tracking capability of the proposed particle filter.