文本文件的语音识别中汉语音节的特征分析

The Features Analysis of Chinese Speech Syllable in Text-file Speech Recognition

汉语语音识别中连续大词汇量的语音识别率较差。若能把连续大词汇量的语音进行实时自动切分为单个音节,便可提高系统的识别率。如何做到对语音识别中音节的自动切分,首先需找出汉语语音音节的特征。本文综合了当前对汉语音节特征的研究成果,通过深入地比较分析,系统地给出了汉语语音音节的功率谱特征和时域特征,为汉语语音音节的自动切分提供算法依据,对提高连续大词汇量语音的识别率有重要意义。

In Chinese speech recognition, the speech recognition ratio of continuous large vocabulary is comparatively poor. If speech sound of continuous large vocabulary can be automatically divided into segments of single syllable, the recognition ratio of system can be raised. How to make automatic segmentation of syllable in the recognition of speech sound? First of all, We should find out the features of Chinese speech syllable. Synthesizing current research achevements of Chinese syllabic feature with thorough comparison and analysis, this paper systematically offers the features of power spectrum and time-domain of Chinese speech syllable, Which provides algorithm basis for automatic segmentarion of Chinese speech syllable and has a great significance for raising the recognition ratio of continuous large vocabulary.