| Comparative Study on Channel Compensation for Robust Speech Recognition |
| |
| 作者姓名： | 赵军辉 匡镜明 黄石磊 |
| |
| 作者单位： | DepartmentofElectronicEngineering,SchoolofInformationScienceandTechnology,BeijingInstituteofTechnology,Beijing100081,China |
| |
| 基金项目： | theNationalNaturalScienceFoundation( 60 3 72 0 89) |
| |
| 摘 要： | Some channel compensation techniques integrated into front-end of speech recognizer for improving channel robustness are described. These techniques include cepstral mean normalization, rasta processing and blind equalization. Two standard channel frequency characteristics, G. 712 and MIRS, are introduced as channel distortion references and a mandarin digit string recognition task is performed for evaluating and comparing the performance of these different methods. The recognition results show that in G. 712 case blind equalization can achieve the best recognition performance while cepstral mean normalization outperforms the other methods in MIRS case which is capable of reaching a word error rate of 3.96 %.
|
| 关 键 词： | 语音识别 比较研究 信道补偿 稳定性 挡板均衡 规范化平均值 |
| 收稿时间： | 2003/6/17 0:00:00 |
Comparative Study on Channel Compensation for Robust Speech Recognition |
| |
| ZHAO Jun-hui,KUANG Jing-ming and HUANG Shi-lei.Comparative Study on Channel Compensation for Robust Speech Recognition[J].Journal of Beijing Institute of Technology,2003,12(4):403-406. |
| |
| Authors: | ZHAO Jun-hui KUANG Jing-ming and HUANG Shi-lei |
| |
| Affiliation: | Department of Electronic Engineering, School of Information Science and Technology, Beijing Institute of Technology, Beijing 100081, China |
| |
| Abstract: | Some channel compensation techniques integrated into front-end of speech recognizer for improving channel robustness are described. These techniques include cepstral mean normalization, rasta processing and blind equalization. Two standard channel frequency characteristics, G.712 and MIRS, are introduced as channel distortion references and a mandarin digit string recognition task is performed for evaluating and comparing the performance of these different methods. The recognition results show that in G.712 case blind equalization can achieve the best recognition performance while cepstral mean normalization outperforms the other methods in MIRS case which is capable of reaching a word error rate of 3.96%. |
| |
| Keywords: | robustness speech recognition cepstral mean normalization rasta processing blind equalization |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《北京理工大学学报(英文版)》浏览原始摘要信息 |
|
点击此处可从《北京理工大学学报(英文版)》下载全文 |
