基于Fisher比的Bark小波包变换的语音特征提取算法

Speech feature extraction algorithm based on the Bark wavelet packet transform with Fisher

为解决MFCC特征参数在噪声环境中识别率低的问题,提出一种基于Fisher比的Bark小波包变换特征提取算法.首先采用小波包变换构造Bark滤波器代替三角形的Mel滤波器.其次采用Fisher对Bark滤波后的特征参数进行选择,去除大量干扰信息,节省特征匹配的时间.仿真实验表明,该算法明显提高系统的识别率和鲁棒性.

In order to solve the problem of low recognition rate of MFCC parameter, a feature extraction algorithm based on the Bark wavelet packet transform with Fisher is put forward. Firstly, wavelet packet transform is used to construct Bark filter, which can replace the trian- gular Mel filter. According to the Fisher criterion, feature parameters filtered by Bark filter are adopted, removing the interference information and saving the time of feature matching. Simu- lation results show that the algorithm is able to enhance the recognition rate and robustness.