论文部分内容阅读
汉语学习日趋广泛,如何快速有效地实现标准汉语语料库供学习者参考并对汉语学习者的发音标准水平做出评测受到了很大程度的关注。该文介绍微软提供的TTS语音引擎,阐述其编程实现过程,利用TTS生成的汉语音频作为标准语料库。采集汉语学习者发音,对TTS生成的汉语音频和汉语学习者的发音作预处理并提取特征值,最后采用DTW(动态时间弯曲)算法实时地进行发音标准水平评测。
Chinese learning is becoming more and more widespread. How to quickly and effectively achieve the standard Chinese corpus for learners ’reference and to evaluate Chinese learners’ pronunciation standards has drawn a great deal of attention. This article introduces the TTS voice engine provided by Microsoft, explains its programming and implementation process, and uses Chinese audio generated by TTS as a standard corpus. The pronunciation of Chinese learners is collected, the Chinese audio generated by TTS and the pronunciation of Chinese learners are preprocessed and their eigenvalues are extracted. Finally, DTW (Dynamic Time Bending) algorithm is used to evaluate pronunciation standard level in real time.