基于CFCC和相位信息的鲁棒性说话人辨识

传统的说话人识别中，人们往往认为人耳对相位信息不敏感而忽略了相位信息对语音识别的影响。为了验证相位信息对说话人识别的影响，提出了一种提取相位特征参数的方法。分别在纯净语音和带噪语音条件下，基于高斯混合模型，通过将相位特征参数与耳蜗倒谱系数（CFCC）相结合，研究了相位信息对说话人辨识性能的影响。实验结果标明：相位信息在说话人识别中也有着重要的作用，将其应用于说话人辨识系统，可明显提高系统的识别率和鲁棒性。

Robust speaker identification based on CFCC and phase information

In the traditional speaker recognition, people tend to think that the human ear is not sensitive to phase information and ignore the influence of phase information on speech recognition. In order to verify the influence of phase information on speaker recognition, an algorithm to extract the phase cepstrum coefficient is proposed in this paper. Respectively, under the condition of pure and noise-corrupted speech, based on Gaussian Mixes Model （GMM）, combining the Cochlear Filter Cepstrum Coefficient （CFCC） with the phase feature parameters, the influence of phase on speaker recognition performance is studied. The experimental results indicate that phase information also plays an important role in speaker recognition and when it is applied to speaker recognition system, the recognition rate and the robustness of system has obvious improvement.