一种语音端点检测算法及其在DSP上的实现

摘要

This paper presents an algorithm for speech endpoint detection based on fuzzy RBF neural network. The algorithm obtains characteristic quantities of speech signals by the wavelet analysis, then feds it into fuzzy RBF neural network to detect speech endpoint, and this algorithm is achieved by TMS320VC5416 DSP as the core circuit. The experiment shows that the end-point detection system has very high accuracy rate. Even in low SNR, speech endpoint can be correctly detected.%提出一种基于模糊RBF神经网络的语音端点检测算法.该算法先利用小波分析提取语音信号的特征量,然后将其输入到模糊RBF神经网络进行端点检测运算,并采用以TMS320VC5416DSP为核心的电路进行算法实现.实验结果表明,该系统的端点检测正确率很高,即使在低信噪比时也能正确地判断语音信号的端点.