| 汉语语音识别的抗噪性前端算法及性能分析 |
| |
| 引用本文： | 林建臻,孙甲松,王作英.汉语语音识别的抗噪性前端算法及性能分析[J].电声技术,2004(3):45-48,52. |
| |
| 作者姓名： | 林建臻 孙甲松 王作英 |
| |
| 作者单位： | 清华大学,电子工程系,北京,100084;清华大学,电子工程系,北京,100084;清华大学,电子工程系,北京,100084 |
| |
| 基金项目： | 犦国家863高技术项目(863-306-ZD03-02-1)，985重大项目(985校-22-攻关-06)，国家自然科学基金项目. |
| |
| 摘 要： | 讨论了欧洲电信标准委员会ETSI提出的分布式语音识别系统的抗噪前端特征提取算法，该算法融合多种抗噪技术。结合汉语语音的特点，进行了汉语语音识别整体框架下的算法实现，并进行了实验和分析，典型噪声环境下的识别结果证明，相对于基线MFCC特征提取算法，稳健性有较大提高。
|
| 关 键 词： | 抗噪 分布式语音识别 前端 汉语语音识别 |
| 文章编号： | 1002-8684(2004)03-0045-04 |
A Noise Robust Front End Algorithm for Mandarin Speech Recognition and Performance Analysis |
| |
| LIN Jian-zhen,SUN Jia-song,WANG Zuo-ying.A Noise Robust Front End Algorithm for Mandarin Speech Recognition and Performance Analysis[J].Audio Engineering,2004(3):45-48,52. |
| |
| Authors: | LIN Jian-zhen SUN Jia-song WANG Zuo-ying |
| |
| Abstract: | Noise robustness of automatic speech recognition system is the hot topic during recent years.In this paper,the noise robust Front End algorithm proposed by ETSI for Distributed Speech Recognition System(DSR)which is a combination of several separate noise-robust techniques is discussed.Considering Mandarin speech char-acters,a feature extraction system based on this algorithm is realized and analyzed its performance through recog-nition experiments is analyzed.In some typical noise environments,we got much higher recognition rate compared with classical MFCC(Melscale Frequency Cepstrum Coeffcient )feature is obtained. |
| |
| Keywords: | noise robustness DSR front end mandarin speech recognition |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
