Research map / Video / Video generation
Echoes Over Time: Unlocking Length Generalization in Video-to-Audio Generation Models
Christian Simon, Masato Ishii, Wei-Yao Wang, Koichi Saito, Akio Hayakawa, Dongseok Shim, Zhi Zhong, Shuyang Cui, Takashi Shibuya, Shusuke Takahashi, Yuki Mitsufuji
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Non-Causal Mamba-2 instead of Transformers without positional embeddings (ablation) | MMHNet-S · UnAV100 | IB-Score: 28.41 -> 36.82 | Improved |
| use Non-Causal Mamba-2 instead of Causal Mamba-2 (ablation) | MMHNet-S · UnAV100 | IB-Score: 33.32 -> 36.82 | Improved |
| use Non-Causal Mamba-2 instead of Transformers (ablation) | MMHNet-S · UnAV100 | DeSync: 0.638 -> 0.439 | Improved |
| use Non-Causal Mamba-2 instead of Causal Mamba-2 (ablation) | MMHNet-S · LongVale | IB-Score: 17.42 -> 30.62 | Improved |
| use Non-Causal Mamba-2 instead of Transformers (ablation) | MMHNet-S · LongVale | IB-Score: 18.65 -> 30.62 | Improved |
| use Non-Causal Mamba-2 instead of Causal Mamba-2 (ablation) | MMHNet-S · UnAV100 | FD_VGG: 2.28 -> 3.35 | Hurt |
| use Non-Causal Mamba-2 instead of Transformers (ablation) | MMHNet-S · LongVale | FD_PANNs: 9.72 -> 10.10 | Hurt |
| add hierarchical routing (ablation) | MMHNet-S · UnAV100 | DeSync: 0.621 -> 0.439 | Improved |
| add hierarchical routing (ablation) | MMHNet-S · LongVale | FD_PASST: 442.12 -> 323.39 | Improved |
| add hierarchical routing (ablation) | MMHNet-S · LongVale | IB-Score: 26.34 -> 30.62 | Improved |
| set token selection threshold to 0.5 instead of 0.3 (ablation) | MMHNet-L · UnAV100 | FD_PANNs: 8.13 -> 5.29 | Improved |
| set token selection threshold to 0.7 instead of 0.5 (ablation) | MMHNet-L · UnAV100 | IB-Score: 36.27 -> 0.02 | Hurt |
Closest papers by meaning
- Diverse and Aligned Audio-to-Video Generation via Text-to-Video Model Adaptation (AAAI 2024 · similarity 0.91)
- MMAudio: Taming Multimodal Joint Training for High-Quality Video-to-Audio Synthesis (CVPR 2025 · similarity 0.90)
- Seeing and Hearing: Open-domain Visual-Audio Generation with Diffusion Latent Aligners (CVPR 2024 · similarity 0.89)
- Aligned Better, Listen Better for Audio-Visual Large Language Models (ICLR 2025 · similarity 0.88)
- AVGen-Bench: A Task-Driven Benchmark for Multi-Granular Evaluation of Text-to-Audio-Video Generation (ICML 2026 · similarity 0.88)
- Foley-Flow: Coordinated Video-to-Audio Generation with Masked Audio-Visual Alignment and Dynamic Conditional Flows (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
