节点文献
运用互信息匹配及关键词分析的语音对话系统
Applying Mutual Information Matching and Keywords Analysis to a Speech Dialogue System
【摘要】 介绍应用于“虎丘”旅游服务的非特定人语音对话系统 .该系统运用互信息匹配模型 MIM进行音节识别 ,并提出了关键词句法模型及相应的分析算法 KBP进行语句分析和识别 .实验表明 ,互信息匹配模型 MIM的应用使连续语音音节识别率达到 78% ,而关键词句法分析的运用使系统总体语句识别率提高了 65 % ,而且对预处理部分音节分割的误差以及不规范的语句输入有较好的容错处理能力
【Abstract】 This paper describes a speech dialogue system for tour service of a famous scenery place "Huqiu" of Suzhou. The system includes three main modules as acoustic phonetic processing, syntactic analysis and semantic extracting. In the system, the Mutual Information Matching model MIM was selected for continuous syllable recognition and a robust
【关键词】 语音对话系统；
互信息；
关键词分析；
KBP；
【Key words】 based syntax analysis algorithm KBP was proposed for sentence parser. The experiments show 78% of test syllables were recognized based on MIM and the application of KBP make whole sentence recognition performance a 65% improvements in average.；
【Key words】 based syntax analysis algorithm KBP was proposed for sentence parser. The experiments show 78% of test syllables were recognized based on MIM and the application of KBP make whole sentence recognition performance a 65% improvements in average.；
【基金】 国家自然科学基金项目 (60 172 0 16)资助;江苏省“计算机信息处理技术”重点实验室开放课题资助项目(KJS0 10 0 2 1)资助
- 【文献出处】 小型微型计算机系统 ,Mini-micro Systems , 编辑部邮箱 ,2003年01期
- 【分类号】TN912.3
- 【被引频次】19
- 【下载频次】214