| 基于音素的非特定人英语命令词识别算法研究 |
| |
| 引用本文： | 贲俊,余小清,万旺根.基于音素的非特定人英语命令词识别算法研究[J].信号处理,2002,18(6):535-538. |
| |
| 作者姓名： | 贲俊 余小清 万旺根 |
| |
| 作者单位： | 上海大学通信与信息工程学院,上海,200072;香港理工大学多媒体创新中心,香港,九龙 |
| |
| 基金项目： | 本文得到国家自然科学基金项目(No.60072031)，香港理工大学研究基金项目(No.G-T499)的资助 |
| |
| 摘 要： | 本文提出了一种新的基于音素的非特定人英语命令词识别算法,并在此算法基础上构建了一个非特定人英语命令词识别系统。结合非特定人语音识别系统的特点,系统的实现采用了HTK与VisualC++两种工具混和使用,提高了整个系统的开发效率。在识别阶段将置信度评估和不完全匹配的方法结合在一起,在一定程度上提高了识别的质量,在词汇量大于10的情况下取得了87.8％的识别率。
|
| 关 键 词： | 语音识别 HTK 置信度 |
| 修稿时间： | 2002年5月20日 |
Research on an Algorithm of Phoneme based Speaker-independent English Commands Recognition |
| |
| Ben Jun Yu Xiaoqing Wan Wanggen.Research on an Algorithm of Phoneme based Speaker-independent English Commands Recognition[J].Signal Processing,2002,18(6):535-538. |
| |
| Authors: | Ben Jun Yu Xiaoqing Wan Wanggen |
| |
| Abstract: | In this paper we propose a new algorithm of phoneme based speaker-independent English commands recognition and develop a speaker-independent English commands recognition system based on it. It can accelerate the whole system development by mixed using HTK and Visual C++ based on character of speaker-independent speech recognition. In recognition phase we combine confidence measures with incomplete matching, which considerably improve the quality of recognition. RA (Recognition accuracy) is 87.8% when the vocabulary is more than 10. |
| |
| Keywords: | Speech recognition HTK Confidence measures |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
