Primary Research On Standard Chinese Visemes

The study of traditional phonetics indicates the shape of lips takes important effect on the articulations of consonants and vowels. AVSP (Audio-Visual Speech Processing) which is increasingly popular in speech-technique field can improve the naturalness of synthetical speech and recognition rate of the speech recognition system. Especially in computer-synthesized face, the variants of lip-shape play a crucial role. The present research aims to theorize a system of lipshape variety comparison of phoneme system of Standard Chinese. A new terminology-iseme system?is given to this system. No one has studied Chinese visemes specially.In this research, a minitype visual speech database was created firstly. A male speaker whose face was marked by painting lip areas and points was recorded uttering 195 monosyllables that were made into image files and sound files at the same time. The images were captured at the rate of 25 frames/sec. According to the spectrogram the frame which represent phoneme is determined. The lips-gesture of phoneme was described and characterized by 11 parameters and the database is made up of 494 records. Then on the basis of the database a viseme modeling of Standard Chinese was ciriven through a series of statistics analysis. The primaiy conclusion is:1.Most of the lip parameters are highly correlated, of which the height of the mouth opening, the protrusion degree of lip corner and the visible degree of upper teeth play a leading role.2.There are eight visemes that are consonants visemes) and vowel visemes) and fourteen variants in Standard Chinese.