SALSA-V: Shortcut-Augmented Long-form Synchronized Audio from Videos
This addresses the need for efficient and high-quality audio synthesis in video applications, such as Foley generation and sound design, though it appears incremental by building on existing multimodal generation approaches.
The authors tackled the problem of generating synchronized, high-fidelity long-form audio from silent videos by proposing SALSA-V, which outperformed state-of-the-art methods in audiovisual alignment and synchronization, achieving rapid generation in as few as eight sampling steps.
We propose SALSA-V, a multimodal video-to-audio generation model capable of synthesizing highly synchronized, high-fidelity long-form audio from silent video content. Our approach introduces a masked diffusion objective, enabling audio-conditioned generation and the seamless synthesis of audio sequences of unconstrained length. Additionally, by integrating a shortcut loss into our training process, we achieve rapid generation of high-quality audio samples in as few as eight sampling steps, paving the way for near-real-time applications without requiring dedicated fine-tuning or retraining. We demonstrate that SALSA-V significantly outperforms existing state-of-the-art methods in both audiovisual alignment and synchronization with video content in quantitative evaluation and a human listening study. Furthermore, our use of random masking during training enables our model to match spectral characteristics of reference audio samples, broadening its applicability to professional audio synthesis tasks such as Foley generation and sound design.