摘要
研究了小波包变换及MFCC参数的提取,结合二者特点,提取了基于小波包变换和MFCC的新参数DWT-MFCC,并在基于16阶GMM系统上进行说话人识别实验。实验结果表明,相对于传统的MFCC参数,在相同的噪声环境下,DWT-MFCC参数具有更高的说话人识别率。
The wavelet packet transform and the parameters are studied. Based on these research, GMM system the speaker recognition is evaluated. accuracy in noisy circnm stances compared with the extraction of the MFCC （Mel-Frequency Cepstral Coefficients） a new parameters, DWT-MFCC, is extracted. Based on the The results show that the DWT-MFCC has higher recognition MFCC parameters.
出处
《电声技术》
2009年第2期49-51,共3页
Audio Engineering
基金
河南省教育厅自然科学研究项目(2008A510013)
许昌学院校内科研基金项目
