Robust sound localization algorithm for intelligent acoustic surveillance system

Abstract

A new sound localization algorithm for intelligent surveillance system is proposed in this paper. This study mainly focuses on the building of the reliable tetrahedron-shaped microphone array that allows noise robust sound localization. The proposed system is derived from the traditional triangular microphone array and developed using a plane coordinate transformation method. A real-time sound localization experiments with a proposed method, the previously known triangular microphone array and existing tetrahedral microphone array system are conducted and compared. The experimental results show that the proposed system outperforms all the comparing systems in terms of the detection rate of the sound source. The proposed system also confirms the robustness of the noise effect in low SNR environment.