V2SFlow: Video-to-Speech Generation with Speech Decomposition and Rectified Flow
This addresses the problem of generating natural speech from unconstrained talking face videos for applications like assistive technology or video editing, representing a strong specific gain rather than incremental.
The paper tackles video-to-speech generation by decomposing speech into content, pitch, and speaker attributes from silent talking face videos, and uses a rectified flow matching decoder to produce speech, achieving results that outperform state-of-the-art methods and even surpass ground truth naturalness.
In this paper, we introduce V2SFlow, a novel Video-to-Speech (V2S) framework designed to generate natural and intelligible speech directly from silent talking face videos. While recent V2S systems have shown promising results on constrained datasets with limited speakers and vocabularies, their performance often degrades on real-world, unconstrained datasets due to the inherent variability and complexity of speech signals. To address these challenges, we decompose the speech signal into manageable subspaces (content, pitch, and speaker information), each representing distinct speech attributes, and predict them directly from the visual input. To generate coherent and realistic speech from these predicted attributes, we employ a rectified flow matching decoder built on a Transformer architecture, which models efficient probabilistic pathways from random noise to the target speech distribution. Extensive experiments demonstrate that V2SFlow significantly outperforms state-of-the-art methods, even surpassing the naturalness of ground truth utterances. Code and models are available at: https://github.com/kaistmm/V2SFlow