Echoes Over Time: Unlocking Length Generalization in Video-to-Audio Generation Models

arXiv:2602.20981v3 Announce Type: replace Abstract: Scaling multimodal alignment between video and audio is challenging, particularly due to limited data and the mismatch between text descriptions and frame-level video information. In this work, we tackle the scaling challenge in multimodal-to-audio generation, examining whether models trained on short instances can generalize to longer ones during testing. To tackle this challenge, we present multimodal hierarchical networks so-called MMHNet, an enhanced extension of state-of-the-art video-to-audio models. Our approach integrates a hierarchical method and non-causal Mamba to support long-form audio generation. Our proposed method significantly improves long audio generation up to more than 5 minutes. We also prove that training short and testing long is possible in the video-to-audio generation tasks without training on the longer durations. We show in our experiments that our proposed method could achieve remarkable results on long-video to audio benchmarks, beating prior works in video-to-audio tasks. Moreover, we showcase our model capability in generating more than 5 minutes, while prior video-to-audio methods fall short in generating with long durations.

Related

• EchoTorrent: Towards Swift, Sustained, and Streaming Multi-Modal Video Generation
• Tora3: Trajectory-Guided Audio-Video Generation with Physical Coherence
• When Numbers Speak: Aligning Textual Numerals and Visual Instances in Text-to-Video Diffusion Models
• Empowering Video Translation using Multimodal Large Language Models

Source: arXiv cs.CV | 2026-04-16