Overview


We propose a flexible framework for joint video-audio generation, visual-steered audio generation, and audio-steered visual (image/video) generation tasks.

Video-to-audio generation


Video credit to OpenAI Sora.

Joint video-audio generation and audio-steered visual generation will come soon.



Stay tuned.

Abstract


Video and audio content creation serves as the core technique for the movie industry and professional users. Recently, existing diffusion-based methods tackle video and audio generation separately, which hinders the technique transfer from academia to industry. In this work, we aim at filling the gap, with a carefully designed optimization-based framework for cross visual-audio and joint visual-audio generation. We observe the powerful generation ability of off-the-shelf video or audio generation models. Thus, instead of training the giant models from scratch, we propose to bridge the existing strong models with a shared latent representation space. Specifically, we propose a multimodality latent aligner with the pre-trained ImageBind model. Our latent aligner shares a similar core as the classifier guidance that guides the diffusion denoising process during inference time. Through carefully designed optimization strategy and loss functions, we show the superior performance of our method on joint video-audio generation, visual-steered audio generation, and audio-steered visual generation tasks.

Research Paper

Supplement Video


Capabilities of our framework and comparisons with baselines.