语音识别中的抗噪声方法*

北京邮电大学学报 ›› 1998, Vol. 21 ›› Issue (2): 10-14.

语音识别中的抗噪声方法^*

崔屹,张东,时良平,陈丽媛

北京邮电大学机械电子工程系, 北京100876

收稿日期:1997-07-22 出版日期:1998-03-10
基金资助:

Robust Noise Suppression Methods in Speech Recognition

Cui Yi, Zhang Dong, Shi Liangping, Chen Liyuan

Department of Mechanical and Electronic Engineering,
Beijing University of Posts and Telecommunications, Beijing 100876

Received:1997-07-22 Online:1998-03-10
Supported by:

摘要/Abstract

摘要： 在实验室环境中正确识别率很高的语音识别系统, 在生产现场中的性能往往会大幅度下降, 严重限制了语音识别技术的实际应用.针对邮政分拣中心的实际生产环境, 对语音识别技术中的抗噪声问题进行了实验分析和研究, 提出了几种在生产现场中具有较好效果的噪声补偿方法, 并给出了部分实验结果.

关键词: 语声识别, 语声信号处理, 抗噪声方法

Abstract: Experiment shows that speech recognition system withhigh recognition rate in laboratory environment will greatly decreased in practical field. This severally restricted the use of speech recognition techniques.We carried out a serials of experiments and studies in the parcel sorting workstation, and proposed robust noise suppression methods which revealed encouragingresults in the practical field. The experimental data are given.

Key words: speech recognition, speech signal processing, noise suppression methods

中图分类号:

TP18

崔屹,张东,时良平,陈丽媛. 语音识别中的抗噪声方法^*[J]. 北京邮电大学学报, 1998, 21(2): 10-14.

Cui Yi, Zhang Dong, Shi Liangping, Chen Liyuan. Robust Noise Suppression Methods in Speech Recognition[J]. JOURNAL OF BEIJING UNIVERSITY OF POSTS AND TELECOM, 1998, 21(2): 10-14.