基于广义加权的AMDF时延估计方法研究

时延估计是声源定位常用的方法。多途效应严重影响声源信号时延估计性能，传统方法难以克服。提出一种基于广义加权的平均幅度差函数（Average Magnitude Difference Function，AMDF）时延估计方法，利用改进的AMDF方法提高对多途效应的抑制，通过广义加权方法降低算法的噪声敏感性。仿真及实验表明，对于窄带信号，该方法能够获得比传统广义互相关方法更高的时延估计性能，估计结果的误差减小，稳定性能提高。

Research on time delay estimation based on generalized weighted AMDF

Time delay estimation (TDE)-based methods are commonly used for localization of sound source. However, multi-path seriously affects the performance of the TDE on the sound source signal. In addition, it is difficult to overcome the effects of multi-path using traditional methods. The TDE method based on generalized weighted average magnitude difference function (Average Magnitude Difference Function, AMDF) was proposed in this study. The multi-path effects can be suppressed by the modified AMDF method and the generalized weighted function can be used to reduce the noise sensitivity of the algorithm. Simulation and experiment results showed that for narrowband signals the performance of the newly proposed method was superior to that of the traditional generalized cross-correlation method. With the proposed method, the error of result was reduced and the stability was considerably improved.