基于河南方音的英语语音自动分析

Automatic Analysis of Henan-accent-dependent English Pronunciation

首先,建立基于方音的二语发音数据库,并分析发音数据的音韵特征;然后,基于隐马尔可夫理论,提出了一种识别特定方音背景的外语学习者的语音产出的可行性方案.性能试验显示,系统可以较为准确地自动识别出以河南方音为背景的学习者的外语语音产出,并形成诊断性评价,进而提升其外语语音水平.

Firstly,the accented second language phonetic corpus is built and its phonological features are analyzed.Then based on HMM,this paper has proposed a feasible scheme for the automatic speech recognition of the foreign language learners＇ speech production from specific accent background in Henan.The result of performance testing shows that the system may recognize Henan-accentedEnglish,evaluate both acceptable and unacceptable productions,and promote their foreign language pronunciation.