Design of an MPEG4-Based Text-Driven 3D Facial Animation System

Abstract

In this paper, we present a text-driven talking head system, which consists of two major components, a text-to-speech (TTS) unit and a facial animation (FA) unit. We adopt a commercial text-to-speech system, Microsoft Speech API, to be our TTS unit and a feature-point-driven facial animation system based on hypothetical face method to be our FA unit. Given meaningful sentences, the text-to-speech system produces speech signals and speech-driven information that includes 2-D lips shape and timing information of the speech. In this paper, the 3-D facial actions of lips for speech using hypothetical face method are derived from the provided 2-D lips shape information and the facial actions of lips are interfaced with speech-driven information in the text-to-speech system in order to generate a more realistic 3-D lip motion animation for speech. Experiment conducted on thus accomplished text-driven 3D talking head system shows that the proposed method is highly acceptable.