分布式语音识别的前端处理及相关标准  

Front end Processing of Distributed Speech Recognition and Related Stan dards

作　　者：王艳琴[1] 梁钊[1] 蒙山[1] 

机构地区：[1]五邑大学信息科学研究所,广东江门529020

出　　处：《电声技术》2002年第5期4-7,共4页Audio Engineering

基　　金：广东省自然科学基金资助项目(000872)

摘　　要：语音识别在实际应用中受到信道噪声和便携终端计算、存储能力不足等因素制约。分布式语音识别(DSR)不仅解决了上述的问题,还有占用带宽窄、综合成本低等优点,但其应用的前提是提取的参数必须标准化。文中介绍了分布式语音识别前端处理的基本结构以及相关标准。In practice,the performance of speech recognition is limited by such factors as channel noise and insufficient computational power and memory size in portable terminals.Dis-tributed speech recognition(DSR)can not only solve those problems but also provides such merits as narrow bandwidth and low cost.However the precondition for its application is the standardiza-tion of the extracted feature parameters.Front end processing of distributed speech recognition and related standards are described in this paper.

关 键 词：分布式语音识别 前端处理 标准 信道噪声 存储 DSR 

分 类 号：TN912.34[电子电信—通信与信息系统]