Estimation of the number of sources and their locations in colored noise using reversible jump MCMC

Abstract

The authors proposed a method of estimating the source locations in colored noise using a hierarchical model in Bayesian estimation framework. In this paper, the reversible jump MCMC method is introduced into this method to jointly estimate the number of sources. By introducing this, the cases in which the number of sources is unknown or dynamically changes can be handled. The results of the experiments show an improvement over the conventional methods in terms of the source localization performance in a room reverberation.