节点文献
利用HMM嵌入训练方法建立汉语电话连续语音声学模型
Modeling For Chinese Continuous Telephone Speech Using HMM’s Embedded-Training Method
【摘要】 文章介绍了用HMM嵌入训练方法来建立连续语音的声学模型,并对基于音素的HMM和基于音节的HMM进行了比较,并以此为基础建立了一个实用的银行电话语音服务系统。
【Abstract】 This paper describes the building of a set of Chinese acoustic models which used to recognize telephone utterances by embedded-training method, and compares phoneme-based HMMs with syllable-based HMMs. Based on these models, a banking service system is constructed.
【关键词】 嵌入训练;
电话语音识别;
音素模型;
音节模型;
银行语音服务系统;
【Key words】 Embedded-training; Telephone speech recognition; Phone-based HMM; Syllable-based HMM; Banking service system;
【Key words】 Embedded-training; Telephone speech recognition; Phone-based HMM; Syllable-based HMM; Banking service system;
- 【文献出处】 计算机工程与应用 ,COMPUTER ENGINEERING AND APPLICATIONS , 编辑部邮箱 ,2000年06期
- 【分类号】TN912
- 【被引频次】7
- 【下载频次】89