小波包变换与Teager能量算子结合的说话人识别

Speaker recognition combining wavelet packet transform with Teager Energy Operator

在说话人识别系统中,语音特征参数的提取是影响系统性能的关键因素之一。在研究了MFCC参数的基础上,结合MFCC参数在信号的低频部分具有高频率分辨率以及小波包变换可以对信号的高频部分进行分解以提高高频部分的频率分辨率的优点,将二者结合,将Teager能量算子引入到信号高频部分的能量参数求解,构造了一种新的混合特征参数,采用支持向量机实现说话人的分类识别。实验结果表明,该特征参数有效提高了说话人辨识系统的识别率。

In speaker recognition system, the key factor is extracting a personality feature of the speaker. Based on analysis of MFCC parameter extraction, this paper constructs a hybrid parameter through combination of the low part of MFCC parameter and using the wavelet packet transform processing the high part of the signal, of which the Teager Energy Operator （TEO） is used. A Support Vector Machine（SVM） is used to do the classification work. Experimental data shows that the method is effective in raising the recognition rate.