Abstract: An improved algorithm for pitch detection is proposed.In this algorithm，noisy speech is enhanced using Least Mean Square（LMS） adaptive filter，and then the Autocorrelation Function（ACF） and the Average Magnitude Difference Function（AMDF） of speech signal are calculated.Through analyzing the peak of ACF and the trough of AMDF，the improved weighted square feature of ACF/AMDF is extracted to detect the speech pitch.Experimental results indicate that the proposed pitch detection algorithm is superior to traditional methods obviously，and has higher detection precision and better robustness in low SNR environments.

Key words: pitch detection, Least Mean Square（LMS） algorithm, Autocorrelation Function（ACF）, Average Magnitude Difference Function（AMDF）

摘要： 提出了一种改进的基音检测算法。利用LMS自适应滤波器对带噪语音信号进行语音增强，再求语音信号的自相关函数（ACF）和平均幅度差函数（AMDF），对ACF峰值和AMDF谷值进行分析，提取改进的ACF/AMDF加权平方特征进行基音检测。实验结果表明，改进算法的检测效果明显优于传统方法，在低信噪比环境下，鲁棒性较好，提高了检测精度。

关键词: 基音检测, LMS算法, 自相关函数, 平均幅度差函数