| 基于MFCC的说话人识别系统 |
| |
| 引用本文： | 郭春霞,裘雪红.基于MFCC的说话人识别系统[J].电子科技,2005(11):53-56. |
| |
| 作者姓名： | 郭春霞 裘雪红 |
| |
| 作者单位： | 西安电子科技大学,计算机学院,陕西,西安,710071 |
| |
| 摘 要： | 说话人识别可以被看作语音识别的一种,是当前的研究热点之一.本文实现的说话人辨认系统,采用能够反映人对语音的感知特性的Mel频率倒谱系数（Mel-Frequency Ceptral Coefficients ,MFCC）作为特征参数.试验比较了MFCC、差分MFCC组合MFCC分别与VQ、DTW相结合的识别率.得出差分MFCC组合MFCC优于MFCC.
|
| 关 键 词： | 说话人识别 Mel频率倒谱系数 特征提取 |
| 收稿时间： | 2005-06-28 |
| 修稿时间： | 2005年6月28日 |
Study of MFCC Speaker Recognition |
| |
| Guo Chunxia,Qiu Xuehong.Study of MFCC Speaker Recognition[J].Electronic Science and Technology,2005(11):53-56. |
| |
| Authors: | Guo Chunxia Qiu Xuehong |
| |
| Abstract: | Speaker recognition can be regarded as a kind of voice recognition. It is one of the current research hotspots. This paper implements the speaker recognition system with Mel Frequency Ceptral Coefficients(MFCC) as its feature parameters. An experimental comparison between the recognition rate of the combination of MFCC with VQ and DTW and that of the combination of difference MFCC combined with MFCCDTW with VQ and DTW indicates that the difference MFCC combined with MFCC is better than MFCC. |
| |
| Keywords: | Speaker identification MFCC feature extraction |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
