面向声源定位的改进广义互相关时延估计方法

面向以数字微机电系统(MEMS)麦克风作为声传感器的声源定位阵列,完成了前端麦克风阵列的电路设计和以现场可编程逻辑门阵列(FPGA)为核心的数据采集系统的开发,并对麦克风输出的脉冲密度调制信号进行了降采样处理.针对传统的广义互相关算法在低信噪比下时延估计误差较大的问题,提出了一种改进PHAT加权函数的方法.在同等条件下对基于不同加权函数的广义互相关算法进行了MATLAB仿真验证,实验结果表明,在低信噪比条件下该方法相较于传统的广义互相关算法,时延估计误差更小且抗噪性能更强.

Improved Generalized Cross-Correlation Time Delay Estimation Method for Sound Source Localization

For the sound source localization array with digital MEMS microphone as the acoustic sensor,the circuit design of the front-end microphone array and the development of the data acquisition system with FPGA as the core are completed, and the down-sampling processing of the pulse density modulation signal output by the microphone is completed. Aiming at the problem that the traditional generalized cross-correlation algorithm has a large time delay estimation error under low signal-to-noise ratio, a method to improve the PHAT weighting function is proposed. Under the same conditions, the generalized cross-correlation algorithm based on different weigh-ting functions is verified by MATLAB simulation. The experimental results show that compared with the traditional generalized cross-correlation algorithm, this method has smaller time delay estimation error and stronger anti-noise performance under low signal-to-noise ratio conditions.