基于改进MFCC和GMM的语音性别识别分析

Analysis of Speech Gender Recognition Based on Improved MFCC and GMM

阐述基于改进MFCC和GMM的语音性别识别技术的特点,通过抽取改进的MFCC语音特征,利用不同阶数、不同协方差矩阵的高斯混合模型来进行说话人性别识别,采用不同参数结果进行对比,得到最优的结果参数。实验结果表明,MFCC和GMM能够有效地用于说话人性别分类。

This paper explains the characteristics of speech gender recognition technology based on improved MFCC and GMM.By extracting improved MFCC speech features,Gaussian mixture models with different orders and covariance matrices are used for speaker gender recognition.Different parameter results are compared to obtain the optimal result parameters.The experimental results indicate that MFCC and GMM can effectively be used for speaker gender classification.