Collision Handling in 3D Articulatory Animation for Chinese Speech Articulation

The aim of this paper is to efficiently handle collisions that happened in the oral cavity. We proposed a collision handling method which includes two stages: collision detection stage for detecting collisions happened among the tongue vs: teeth and the tongue vs: palate; and collision response stage which is used to avoid a physically impossible situation. Here, the physically impossible situation indicates that the virtual tongue would penetrate the rest articulators in the oral cavity during its movements. The movements of the virtual tongue are driven by the data recorded by three electromagnetic articulography (EMA) sensors attached to a female Chinese speaker's tongue when the desired Chinese speech articulations are being articulated. Here, two Chinese speech articulations are used in our experiment to test our method: /ba/ and /t'a/.