Abstract: This paper constructs a double modal voice and shape matching control model driven by the text and reading voice. With the combination of the double driven mode，the information insufficiency in single driven mode is counteracted and complemented. During the disposal of matching, the accurate mouth viseme is calculated with the spelling of text, and the correct time sequence of mouth viseme is given by read-speech. Therefore, the mouth motion for read-speech can be matched suitably and correctly.

Key words: matching between speech and mouth shape, double modal voice, mouth viseme