Abstract: Speech source localization technology based on MUSIC algorithm has high estimation accuracy and stability, and the computational complexity of the algorithm is mainly concentrated in the process of the spectrum searching. Search space of spectrum can be reduced by the estimation of azimuth through the delay estimation speech source localization algorithm. As a result, the amount of computation of the MUSIC algorithm can be reduced. Experimental results show that this method can drastically reduce the amount of computation, and ensure the accuracy of localization. In the same time, this method is valuable for engineering applications.

Key words: speech source localization, MUSIC algorithm, angle pre-estimation, spectrum searching

摘要： 基于MUSIC算法的声源定位技术具有很高的估计精度及稳定性，算法的运算量主要集中在谱搜索部分。可以通过时延估计声源定位算法预先估计出方位角来缩小谱搜索空间，降低MUSIC算法的运算量。实验结果表明，该方法在保证定位精度的同时，可以大大地降低运算量，对工程应用具有一定的参考价值。

关键词: 声源定位, MUSIC算法, 角度预估计, 谱搜索