Speech Source Tracking Based on Distributed Particle Filter in Reverberant Environments

Abstract

In reverberant and noisy environments, tracking a speech source in distributed microphone networks is a challenging problem. A speech source tracking method based on distributed particle filter (DPF) and average consensus algorithm (ACA) is proposed in distributed microphone networks. The generalized cross-correlation (GCC) function is used to approximate the time difference of arrival (TDOA) of speech signals received by two microphones at each node. Next, the multiple-hypothesis model based on multiple TDOAs is calculated as the local likelihood function of the DPF. Finally, the ACA is applied to fuse local state estimates from local particle filter (PF) to obtain a global consensus estimate of the speech source at each node. The proposed method can accurately track moving speech source in reverberant and noisy environments with distributed microphone networks, and it is robust against the node failures. Simulation results reveal the validity of the proposed method.