蒙古语主要方言的声学和音律特征分析分类

Aanlyzing and Classifying Mongolian Major Dialects by Acoustic and Prosodic Features

本文通过语言的声学特征、音律信息及语言结构的统计概率考察了蒙古语的 主要方言。分析方法包括：语音的 ＬＰＣ特征参数共振峰频率分析；方言语音识别分 析和识别音素串（口语）与文语（书面语）比较分析；利用方言对照字典，语言结构概 率统计分析；利用逐渐逼近法分段线近似基音（Ｆ０）曲线，抽出各分段线的起始频率 和斜率参数后，用判别分析法进行方言分析。分析结果表明，上述各类方法不仅直 观、符合实际而且信赖性较好。

This paper reports the results of analyzing and classifying the Mongolian major dialects, i. e., Khalha(Mongolia), Chakhar (Inner Mongolian, China) and Oirat (Khalmakh, Xin Jiang, China), based on acoustic and prosodic features. The analyzing methods were included: (l )The formant frequency analysis of vowels. (2) Comparing analysis of the statistical Probabilities of phonemes and syllables in the matching dictionary for dialects. (3 ) Recognition analysis of dialectal speech and comparing analysis spoken from the recognized strings to whiten. (4) Dialect classification based on the fundamental frequency (F0 ) contours of speech. The fundamental frequency contours were approximated by segmented lines so that the mean square error between the lines and F0 values was minimized. The starting frequency and slope were calculated from the segmented lines and used for the analysis. Results show that methods are intuitive and effective.