| 基于STRAIGHT谱的非特定人数字语音识别 |
| |
| 引用本文： | 姚建霄,张歆奕.基于STRAIGHT谱的非特定人数字语音识别[J].五邑大学学报(自然科学版),2011,25(1):56-60. |
| |
| 作者姓名： | 姚建霄 张歆奕 |
| |
| 作者单位： | 五邑大学信息工程学院,广东江门,529020 |
| |
| 摘 要： | 介绍了STRAIGHT算法的原理，并选择STRAIGTH谱作为语音识别的特征参数．采用对应点映射方法以提高同一发音不同样本参数间的匹配效果，在特征匹配的方法上选择差别子空间法，将二者结合应用于非特定人汉语数字0～9的语音识别，实验结果表明．基于STRAIGHT谱的非特定人数字语音识别可以达到97％的识别率．
|
| 关 键 词： | 语音识别 STRAIGHT谱 差别子空间 对应点映射 |
Research on STRAIGHT Spectrum in Speech Recognition |
| |
| YAO Jian-xiao,ZHANG Xin-yi.Research on STRAIGHT Spectrum in Speech Recognition[J].Journal of Wuyi University(Natural Science Edition),2011,25(1):56-60. |
| |
| Authors: | YAO Jian-xiao ZHANG Xin-yi |
| |
| Institution: | YAO Jian-xiao,ZHANG Xin-yi(School of Information Engineering,Wuyi University,Jiangmen 529020,China) |
| |
| Abstract: | The principle of the STRAIGHT algorithm was introduced and the STRAGHT spectrum as feature index for speech recognition was chose.Corresponding point mapping was used to improve the effect of matching different sample parameters and the difference subspace was used for feature matching.These methods can be applied to digital identification of non-specific persons.The result shows that the STRAIGHT spectrum-based digital voice recognition can achieve a high recognition rate. |
| |
| Keywords: | speech recognition STRAIGHT spectrum difference subspace corresponding points mapping |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
