An efficient parametric model for real-time 3D tongue skeletal animation

Abstract

In this paper, we propose a very efficient novel parametric model to describe the surface and structure of the human tongue and a corresponding mathematical model for performing 3D tongue animation. A skeletal chain of virtual bones is automatically generated depending on the geometric features of the 3D object, allowing each tongue segment to be easily manipulated by its corresponding parameters, thus leading to a very eye-pleasant and anatomically correct tongue animation. We also provide a very useful non-linear falloff method for defining the bone influence over its surrounding vertices and a novel mathematical model for performing the skeletal chain transformations with minimal tongue volume variation. Tongue animation is a key task for all visible speech researchers and is highly helpful to describe speech production and the articulation of phonemes.Keywords -3D tongue animation, skeletal animation, 3D tongue parametric model, computational geometry I.

3D Visual Speech Animation from Image Sequences

“…Approaches for 3D tongue animation for visual speech include skeletal animation or morph-target animation. The desired parametric trajectories in both cases can be derived by using parametric keyframing or using EMA data as tongue motion data analogous to that in concatenative speech synthesis [25,14,12,17,6]. Only a few of these approaches take care of aspects like volume preservation and penetration avoidance [27,22].…”