DiVISe: Direct Visual-Input Speech Synthesis Preserving Speaker Characteristics And Intelligibility
This work addresses the problem of generating natural speech from video for applications like assistive technologies, though it is incremental as it builds on existing pre-training methods.
The paper tackles the challenge of video-to-speech synthesis by introducing DiVISe, an end-to-end model that generates speech from silent video without acoustic hints, achieving superior performance in both speaker characteristics preservation and acoustic intelligibility on LRS2 and LRS3 datasets.
Video-to-speech (V2S) synthesis, the task of generating speech directly from silent video input, is inherently more challenging than other speech synthesis tasks due to the need to accurately reconstruct both speech content and speaker characteristics from visual cues alone. Recently, audio-visual pre-training has eliminated the need for additional acoustic hints in V2S, which previous methods often relied on to ensure training convergence. However, even with pre-training, existing methods continue to face challenges in achieving a balance between acoustic intelligibility and the preservation of speaker-specific characteristics. We analyzed this limitation and were motivated to introduce DiVISe (Direct Visual-Input Speech Synthesis), an end-to-end V2S model that predicts Mel-spectrograms directly from video frames alone. Despite not taking any acoustic hints, DiVISe effectively preserves speaker characteristics in the generated audio, and achieves superior performance on both objective and subjective metrics across the LRS2 and LRS3 datasets. Our results demonstrate that DiVISe not only outperforms existing V2S models in acoustic intelligibility but also scales more effectively with increased data and model parameters. Code and weights can be found at https://github.com/PussyCat0700/DiVISe.