基于麦克风阵列的声源定位算法研究及DSP实现

摘要

In all current sound source localization methods, the method based on time delay estimation takes the least computation and is the most easy real time application.The accuracy and robustness of time delay estimation are of critical importance to realize the accuracy and stability of the system which is sound source localization by microphone.However, there are always greater environmental noise and strong reflection in the actual environment, and they make an impact on time delay estimation accuracy, further influence the accurate estimates of sound source position.In order to improve the positioning accuracy, the algorithm of GCC-PHAT-pycombined with the smooth of multiple frames weighted is applied to estimate time delay.The experiment provides the algorithm still has higher accuracy and stronger real-time capability in noisy environment.Further more, a four-channel real-time sound source localization system is designed and implemented,and the performance of the system is tested.The results of experiment show, the accuracy of angle could be controlled within a smaller range in a noisy and reverberant laboratory, and the system is of practical and stability in usage.%基于时延估计的声源定位技术是目前出现的定位方法中计算量最小、最易于实时应用的方法.时延估计的精度和鲁棒性是实现麦克风阵列声源定位系统精确与稳定的关键要素.然而在实际环境中大多存在较大的环境噪声和比较强的反射,影响了时延估计的准确性,进而影响对声源位置的准确估计.为了提高定位的精度,本文采用结合多帧加权平滑方法的GCC-PHAT-ρy进行时延估计,仿真证明这种时延估计的方法在有噪声和混响存在的条件下仍然有比较高的准确性和比较好的实时性.搭建了一个四通道声源实时定位系统,对该系统的性能进行了测试.实验结果表明,在实验室的有噪音和反射的环境中,该系统的角度定位精度基本上能够控制在一个较小的范围内,具有比较强的应用性和稳定性.