语音识别在英语发音纠错中的应用

摘要

主要研究如何更好地让计算机智能地纠正英语学习者的发音错误.借助语音识别中的HMM(隐马尔可夫模型)建模方法,用Viterbi算法和改进的后验概率算法对中国学习者的英语发音进行自动识别,通过对基本单元进行切分和评分,最后,为英语学习者提供可信度比较高的发音信息反馈,纠正发音错误.%The main research target in this paper focuses on better correcting intellectually by the computer the pronunciation errors of the English learners.The system uses the HMM modelling method in ASR (automatic speech recognition) ,the Viterbi algorithm and the improved posterior probability algorithm to automatically recognise the English pronunciation of the Chinese speaker, by segmenting and scoring basic unit;in the end it provides a high believable feedback of pronunciation information to the English learners, and corrects the mistakes in pronunciation.