Frieren: Efficient Video-to-Audio Generation Network with Rectified Flow Matching

TL;DR

Frieren is a novel video-to-audio generation model utilizing rectified flow matching, achieving high-quality, synchronized audio synthesis efficiently with fewer sampling steps, outperforming previous methods.

Contribution

The paper introduces Frieren, a rectified flow matching-based V2A model that improves audio quality, synchronization, and efficiency over existing autoregressive and diffusion models.

Findings

Achieves state-of-the-art quality and synchronization on VGGSound dataset.

Reaches 97.22% alignment accuracy.

Improves inception score by 6.2% over diffusion baselines.

Abstract

Video-to-audio (V2A) generation aims to synthesize content-matching audio from silent video, and it remains challenging to build V2A models with high generation quality, efficiency, and visual-audio temporal synchrony. We propose Frieren, a V2A model based on rectified flow matching. Frieren regresses the conditional transport vector field from noise to spectrogram latent with straight paths and conducts sampling by solving ODE, outperforming autoregressive and score-based models in terms of audio quality. By employing a non-autoregressive vector field estimator based on a feed-forward transformer and channel-level cross-modal feature fusion with strong temporal alignment, our model generates audio that is highly synchronized with the input video. Furthermore, through reflow and one-step distillation with guided vector field, our model can generate decent audio in a few, or even only one sampling step. Experiments indicate that Frieren achieves state-of-the-art performance in both generation quality and temporal alignment on VGGSound, with alignment accuracy reaching 97.22%, and 6.2% improvement in inception score over the strong diffusion-based baseline. Audio samples are available at http://frieren-v2a.github.io.