Speech-Driven Expressive Talking Lips with Conditional Sequential Generative Adversarial Networks

TL;DR

This paper introduces a speech-driven conditional GAN model that generates realistic expressive lip movements for virtual agents, effectively modeling emotion and lexical content without needing transcripts, and outperforms existing methods.

Contribution

It proposes a novel conditional sequential GAN (CSG) framework that models emotion and speech content interactions directly from speech features, improving lip movement realism.

Findings

CSG outperforms three state-of-the-art baselines objectively and subjectively.

Emotion-dependent models enhance lip movement expressiveness.

Emotion-aware adaptation improves emotional consistency in generated movements.

Abstract

Articulation, emotion, and personality play strong roles in the orofacial movements. To improve the naturalness and expressiveness of virtual agents (VAs), it is important that we carefully model the complex interplay between these factors. This paper proposes a conditional generative adversarial network, called conditional sequential GAN (CSG), which learns the relationship between emotion and lexical content in a principled manner. This model uses a set of articulatory and emotional features directly extracted from the speech signal as conditioning inputs, generating realistic movements. A key feature of the approach is that it is a speech-driven framework that does not require transcripts. Our experiments show the superiority of this model over three state-of-the-art baselines in terms of objective and subjective evaluations. When the target emotion is known, we propose to create emotionally dependent models by either adapting the base model with the target emotional data (CSG-Emo-Adapted), or adding emotional conditions as the input of the model (CSG-Emo-Aware). Objective evaluations of these models show improvements for the CSG-Emo-Adapted compared with the CSG model, as the trajectory sequences are closer to the original sequences. Subjective evaluations show significantly better results for this model compared with the CSG model when the target emotion is happiness.