Real-time sound source localization using hybrid framework

Abstract

Real-time sound source localization with microphone array still remains as a difficult task. The steered response power–phase transform (SRP–PHAT) method is proved to be robust, but it suffers from its high computational cost. In this paper, we propose a hybrid framework for real-time sound source localization with microphone array. The main concept is using the results of generalized cross correlation (GCC) based time difference of arrival (TDOA) estimation to narrow down the search space of SRP–PHAT. A circular clustering algorithm is developed to interlink the above two parts and calculate the extracted search space. The proposed hybrid method maintains the high localization accuracy of traditional SRP–PHAT algorithm, and the computational cost is much reduced. Moreover, any existing improvement of SRP algorithm can be directly incorporated into the hybrid framework to further improve performance. Experiment and simulation results validate the efficiency and accuracy of the proposed method.