WANG Guirong,JIN Xiaofeng*.Contrastive analysis of pure vowels probability distributionof Korean and Mongolian in speech segments[J].Journal of Yanbian University,2018,44(03):260-265.
语音段中朝蒙单元音概率分布的对比分析
- Title:
- Contrastive analysis of pure vowels probability distribution of Korean and Mongolian in speech segments
- Keywords:
- speech information processing; acoustic phonetics; contrastive analysis of phonetic; measurement of phonetic similarity; silence segment detection
- 分类号:
- TP391.42
- 文献标志码:
- A
- 摘要:
- 为了从声学语音学角度比对朝鲜语与蒙古语的相似性,提出了一种利用相似单元音在连续语音段中出现的概率分布来对比分析语音相似程度的方法.首先,采用双门限静音段检测方法将连续语音自动划分为一系列的语音段,然后采用模式识别方法检测和统计5对相似单元音在语音段首、段中和段尾的概率分布,最后通过对比分析概率分布比较了朝鲜语和蒙古语的语音相似性.对比分析结果表明,朝鲜语与蒙古语语音之间存在差异性,这两种语音的声学特征不具有相似性.本文研究方法和结果可为进一步研究朝鲜语与蒙古语的语音对比提供参考.
- Abstract:
- In order to compare the similarity between Korean and Mongolian from the perspective of acoustic phonetics, a method for comparing and analyzing the degree of similarity of speech using a probability distribution of similar pure-vowels in continuous speech segments is proposed. First, the double-threshold silence segment detection method is used to automatically divide continuous speech into a series of speech segments. Then, the pattern recognition method is used to detect the probability distribution of 5 pairs of similar pure-vowels statistics in the head, middle and end segments of the segmented speech segment, and finally through comparison. Analyzing the probability distribution compares the similarity of Korean and Mongolian speech. The results of comparative analysis show that there are differences between Korean and Mongolian speech, indicating that the acoustic characteristics of these two speeches are not similar. The research methods and results in this paper can provide references for further study of the phonetic comparison between Korean and Mongolian.
参考文献/References:
[1] 刘双君,金小峰,崔荣一.基于基频的朝鲜语方言辨识方法的研究[J].中文信息学报,2017,31(2):55-60.
[2] 朴明姬,芦世丹,崔荣一.多位朝鲜语数字语音识别方法的研究[J].延边大学学报(自然科学版),2011,37(4):337-341.
[3] 德格吉呼,巴图格日勒,金玲.保安语、东乡语、蒙古语单音节短元音声学对比分析[J].丝绸之路,2015,297(8):54-55.
[4] 徐荣.汉韩元音对比与韩语语音教学[J].高教学刊,2017,66(17):81-83.
[5] 许闰星.朝鲜语-蒙古语单元音和韵律特征的实验语音学对比研究[D].延吉:延边大学,2013:42-68.
[6] Juneja A, Espy-Wilson C. Speech segmentation using probabilistic phonetic feature hierarchy and support vector machines[C]//International Joint Conference on Neural Networks. Killarney, Ireland, 2003:675-679.
[7] 吕卫强,黄荔.基于短时能量加过零率的实时语音端点检测方法[J].兵工自动化,2009,28(9):69-70.
[8] 薛胜尧.基于改进型双门限语音端点检测算法的研究[J].电子设计工程,2015,23(4):78-81.
[9] Mporas I, Ganchev T, Fakotakis N. Speech segmentation using regression fusion of boundary predictions[J]. Computer Speech & Language, 2010,24(2):273-288.
[10] Ranjard L, Reed B S, Landers T J, et al. MatlabHTK: a simple interface for bioacoustic analyses using hidden Markov models[J]. Methods in Ecology & Evolution, 2017,8(5):8-12.
备注/Memo
收稿日期: 2018-05-12 *通信作者: 金小峰(1970—),男,教授,研究方向为智能信息处理.
基金项目: 延边大学一流学科建设项目(18YLPY14); 吉林省科技厅自然科学基金资助项目(20140101225JC)