摘要： 为提高语音识别的准确率，采用文本分析方法，选择发音时容易发错的一系列汉语文本语料，以新疆第一批双语班的维吾尔族男女生为发音人进行录制。利用讯飞普通话标准男女音进行实验比较，从声学角度出发，分析维吾尔人的汉语发音特点，并总结发音规律。通过测量音长、音强、共振峰等证明发音偏误的存在及其原因，可用于语音识别的研究。

关键词: 声学特征参数, 音强, 音长, 共振峰, 发音特点

Abstract: In order to improve the accuracy of the speech recognition, this paper starts from text analysis module and chooses some words, with easy wrong pronunciation, for the Chinese text corpus. These words are recorded by one male and one female speaker who are the first batch of Xinjiang Uyghur students graduated from the bilingual class. The comparison of experimental is done using the IFLYTEK mandarin standard male and female voice. The features of Chinese pronunciations of Uyghur people are analyzed from the acoustic point of view, and some of the rules are summarized. By measuring the acoustic parameters, the presence of errors in pronunciation is confirmed and the reasons are analyzed. Experimental basis for the research of speech recognition is provided.

Key words: acoustic feature parameter, sound intensity, sound duration, formant, pronunciation feature

中图分类号: 

• TP391