Abstract: In this paper sound pressure is sampled through spherical microphone array. Sound field can be decomposed by spherical harmonics and a signal model is built. Based on the model, MUSIC algorithm is adopted to estimate the positions of multiple acoustic sources. However this method fails when the signals are coherent, especially near to each other. Therefore a new petal-blocking method is proposed based on spatial smoothing to improve localization effect. A 72-element spherical microphone array is applied to the simulation. The results show that this method can localize multiple coherent sources and resist noise.

Key words: spherical microphone array, spherical harmonics, Multiple Signal Classification（MUSIC） algorithm, spatial smoothing

摘要： 通过球麦克风阵列采集高阶声场的声压信息，采用球谐函数分解声场并建立信号模型，应用MUSIC算法提取出声源的方位。由于MUSIC算法在信号源相干性比较高，特别是声源比较接近时，其分辨率会严重下降，提出了一种基于空间平滑的瓣分块方法来提高定位的效果。仿真实验采用了一个72元的球阵，实验结果表明：提出的方法能同时有效地确定声场中多个声源的位置，能比较好地对抗噪声的影响。

关键词: 球麦克风阵列, 球谐函数, 多信号分类（MUSIC）算法, 空间平滑