Abstract: Head-worn microphone array has practical applications in man-portable anti-sniper detection positioning systems and robotic acoustic localization system. Usually, the sound source localization method is based on a linear or nonlinear unblocked microphone array. In this paper, considering the diffraction effect of low frequency for back microphones due to the helmet blocking, it uses a head-worn microphone array to design and research the sound source localization algorithm for the sound signal in low frequency band. This method uses the diffraction path of low-frequency acoustic waves to calculate the time delay, and utilizes the Steered Response Power and the Phase Transform（SRP-PHAT） framework for time delay compensation search positioning. Experimental results show that compared to the common unblocked localization algorithm, the method for a head-worn microphone array based on the diffraction path uses the data of microphones back to source. Obviously, it improves the accuracy of localization. Besides, the improvement achieves the best result within 1 kHz frequency band.

Key words: head-worn microphone array, diffraction path, Steered Response Power and the Phase Transform（SRP-PHAT）, sound localization

摘要： 头佩式麦克风阵列在单兵便携反狙击声探测定位系统和机器人声定位系统中具有实际的应用价值。一般的声源定位方法是基于无遮挡的线性或非线性麦克风阵列。采用头佩式麦克风阵列，考虑到背向声源麦克风的低频声波由于头盔遮挡而发生的衍射作用，针对低频波段的声音信号进行定位算法的设计和研究。该算法利用低频声波的绕射路径计算时延，采用联合可控功率响应（SRP-PHAT）框架进行时延补偿搜索定位。实验表明，相比于普通的无遮挡定位算法，基于绕射路径的头佩式麦克风阵列定位方法通过综合利用背向声源的麦克风数据，明显地提高了定位的精度，这种精度的提升在选择1 kHz以内的信号频率窗口时达到最佳效果。

关键词: 头佩式麦克风阵列, 绕射路径, 联合可控功率响应框架（SRP-PHAT）, 声源定位