The development direction of English phonetics intelligent recognition system towards the on-line and continuous human-computer interaction,therefore an on-line continuous interactive English phonetics intelligent recognition system was de-signed. The voice recognition function based on Mel frequency cepstrum coefficient was designed. The NOSE algorithm is selected in system reference database to calculate the voice evaluation score and check the error. The expert knowledge base is used to correct the English phonetics of users. The system is composed of the speech training module and on-line continuous interactive module. The former provides the corpus and phonetic resources of English phonetics for the system database. The latter performs the evaluation and correction for the system database. The experimental results show that the system has high speech recognition rate,short recognition time,and high effectiveness.%英语语音智能识别系统的发展方向是在线、连续性的人机交互,为此,设计在线连续交互式英语语音智能识别系统,并基于梅尔频率倒谱系数设计语音识别函数.系统的参考数据库选取NOSE算法计算语音评价分数并查错,利用专家知识库纠正用户英语语音.系统由语音训练模块和在线连续交互式识别模块组成,前者为系统数据库提供英语语音的语料、音标资源,后者实施系统数据库的评价、纠正工作.实验结果证明设计的系统语音识别率高且识别时间短,有效性强.
展开▼
