The Acoustic Analysis Of Labial In Tibetan Xiahe Dialect

With the development of multimedia technology, researches on the interaction of different media attract more and more attention. Research on visual speech is an important branch of this domain, and has an important role in intelligent human-computer interface. Talking into consideration the characteristics of Chinese language, this thesis systematically studies the problem of automatic viseme parameter estimation, viseme modeling under different descriptions, and acoustic-visual mapping. The main divisions of this thesis are as follows:First, the paper according to the Tibetan Xiahe dialect pronunciation characteristics and the pronunciation of the characters combined with the previous academic research, designs lip annotation platforms for Tibetan Xiahe dialect. The platform can read AVI format video files, and makes a tagging for the video, then extraction the relevant parameters of the lip;Secondly, clustering analysis to made of vowels in Tibetan Xiahe dialect, and the vowel static vision is obtained of Tibetan Xiahe dialect. Combined with the traditional phonetic research method, description is done of static lip positions;Finally, according to the lip annotation platforms, mark the lip of consonants, extraction of the outer lip width, inner lip opening degree and other related parameters. Based on linguistic theory and combination of these parameters, contrast the outer lip width and the inner lip opening in Tibetan Xiahe dialect. The features of lip have taken form in Tibetan Xiahe dialect.This paper based on the characteristics of Tibetan speech, make description of the lip shape of Xiahe dialect and gets some distinctive features, and provides some useful information for Tibetan speech research.