基于元音声学空间的汉语方言与普通话间话者识别探究

An Empirical Study on Forensic Speaker Identification between Chinese Dialect and Mandarin Based on Vowel Acoustic Space

目的旨在为解决汉语方言与普通话之间的方言性伪装语音话者识别可供比对的相同音段数量不足等问题。方法以元音为比对对象,通过将说话人普通话和方言的元音放在同一元音声学空间,统计两者之间相同元音和不同元音的特征异同。结果在元音声学空间中,综合考察元音空间轮廓、元音相似度、元音位置关系以及鼻化现象差异性四个指标,说话人表现出较好的个体稳定性和个体间差异性。结论元音声学空间是当前有效解决方言性伪装语音话者识别的途径之一,也是有效解决司法话者识别中可供比对相同音段数量较少的方法之一。

Objective To solve the problem of insufficient number of identical phonetic segments available for comparison between Chinese dialects and Mandarin for the identification of dialectal disguised speech speaker.Method By putting the vowels of speakers’Mandarin and dialect into the same vowel acoustic space,the similarities and differences of the same vowel and different vowels between them were counted.Results In the vowel acoustic space,the speakers showed good individual stability and inter-individual variability when the four indicators,vowel space contour,vowel similarity,vowel position relationship,and variability of nasalization phenomenon,were examined together.Conclusion The vowel acoustic space is one of the current ways to effectively solve the dialectal disguised speech speaker identification,and also one of the methods to effectively solve the low number of valid phonetic segments available for comparison in forensic speaker identification.