Sound Source Localization Based On Microphone Array Research

Acoustic source localization based on microphone array can be used to determine and track an active talker automatically via picking up acoustic signal by microphone array. It is a hotspot in acoustic signal processing field and has wide applied foreground and practical signification. Focus on this problem, the following work has been done:Main acoustic source localization methods based on microphone array have been summarized and a realization mode of room impulse response in room reverberation environment is proposed. Wide band MUSIC acoustic localization algorithm is presented based on the near field signal mode. Time Delay Estimation (TDE) algorithms are researched, including General Cross Correlation (GCC) algorithm and LMS adaptive algorithm. A new modified Cross-power Spectrum Phase (CSP) GCC algorithm based Voice Activity Detection (VAD) is proposed. It has the advantage of lower computational complexity and better performance. Another new LMS adaptive CSP-GCC algorithm is proposed. Localization algorithms based on TDE are researched, including Maximum Likelihood (ML) and Spherical Interpolation (SI). A new closed-form One Step Least Square (OSLS) localization algorithm based on Eigenvalue Decomposition is proposed. It is mathematically equivalent to SI but with less computational complexity. A practical acoustic source localization system based on TDE is proposed. It has two modules, including modified CSP-GCC TDE based on VAD and OSLS localization. The system has good performance and less computational complexity and can be used in practical environment.