随机轨迹模型中汉语连续语音识别基元的选取方案  

Choice of Acoustic Units for Continuous Chinese Speech Recognition Based on Stochastic Trajectory Models

作　　者：黄心晔 施嵘 

机构地区：东南大学无线电工程系

出　　处：《东南大学学报（自然科学版）》1999年第4期48-51,共4页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金

摘　　要：在比较各种汉语语音识别基元的基础上，结合汉语语音和随机轨迹模型的特色，提出了以音素类单元作为汉语连续语音识别系统的识别基元，基于音素基随机轨迹模型的汉语连续语音识别系统的实验结果表明，该方案可行。After analyzing the characteristics of Chinese speech, the acoustic units for continuous Chinese speech recognition based on stochastic trajectory models are discussed and phoneme like unites are suggested. The experimental results of continuous Chinese speech recognition system based on stochastic trajectory models demonstrate the efficiency of phoneme like units.

关 键 词：识别基元 随机轨迹模型 汉语 语音识别 

分 类 号：TN912.34[电子电信—通信与信息系统] TP391.42[电子电信—信息与通信工程]