汉语语音识别的抗噪性前端算法及性能分析  

A Noise Robust Front End Algorithm for Mandarin Speech Recognition and Performance Analysis

作　　者：林建臻[1] 孙甲松[1] 王作英[1] 

机构地区：[1]清华大学电子工程系,北京100084

出　　处：《电声技术》2004年第3期45-48,52,共5页Audio Engineering

基　　金：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。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.[

关 键 词：汉语 语音识别 抗噪性 性能分析 抗噪前端 特征提取算法 

分 类 号：TN912.34[电子电信—通信与信息系统]