Abstract: In this paper, we propose a complete pipeline of efficient and low-cost techniques to construct a realistic 3D text-driven emotive audio-visual avatar from a single 2D frontal-view face ...