Unified speech and gesture synthesis using flow matching
This addresses the problem of generating realistic multimodal communicative behavior for applications like virtual agents or animation, representing a novel method for a known bottleneck rather than incremental.
The paper tackles joint synthesis of speech acoustics and 3D gestures from text using a unified architecture with optimal-transport conditional flow matching, resulting in improved naturalness, human-likeness, and cross-modal appropriateness compared to benchmarks.
As text-to-speech technologies achieve remarkable naturalness in read-aloud tasks, there is growing interest in multimodal synthesis of verbal and non-verbal communicative behaviour, such as spontaneous speech and associated body gestures. This paper presents a novel, unified architecture for jointly synthesising speech acoustics and skeleton-based 3D gesture motion from text, trained using optimal-transport conditional flow matching (OT-CFM). The proposed architecture is simpler than the previous state of the art, has a smaller memory footprint, and can capture the joint distribution of speech and gestures, generating both modalities together in one single process. The new training regime, meanwhile, enables better synthesis quality in much fewer steps (network evaluations) than before. Uni- and multimodal subjective tests demonstrate improved speech naturalness, gesture human-likeness, and cross-modal appropriateness compared to existing benchmarks. Please see https://shivammehta25.github.io/Match-TTSG/ for video examples and code.