Research map / Video / Video generation
AV-Link: Temporally-Aligned Diffusion Features for Cross-Modal Audio-Video Generation
Moayed Haji-Ali, Willi Menapace, Aliaksandr Siarohin, Ivan Skorokhodov, Alper Canberk, Kwot Sin Lee, Vicente Ordonez, Sergey Tulyakov
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 |
|---|---|---|---|
| fix conditioning flow time instead of uniform sampling (ablation) | AV-Link V2A · VGGSounds V2A ablation | FAD: 6.91 -> 4.79 | Improved |
| fix conditioning flow time instead of uniform sampling (ablation) | AV-Link V2A · VGGSounds V2A ablation | FD: 27.33 -> 18.91 | Improved |
| use frozen diffusion features instead of CLIP features (ablation) | AV-Link V2A · VGGSounds V2A ablation | Onset ACC: 0.386 -> 0.415 | Improved |
| use frozen diffusion features instead of CLIP features (ablation) | AV-Link V2A · VGGSounds V2A ablation | FAD: 2.49 -> 4.79 | Hurt |
| use frozen diffusion features instead of CAVP features (ablation) | AV-Link V2A · Movie Gen benchmark | IS: 4.62 -> 6.53 | Improved |
| add symmetric feature reinjection to Fusion Blocks (ablation) | AV-Link V2A · VGGSounds V2A ablation | FAD: 6.60 -> 4.79 | Improved |
| add symmetric feature reinjection to Fusion Blocks (ablation) | AV-Link V2A · VGGSounds V2A ablation | Onset ACC: 0.365 -> 0.415 | Improved |
| use Fusion Blocks instead of direct alignment (ablation) | AV-Link V2A · VGGSounds V2A ablation | Onset ACC: 0.283 -> 0.415 | Improved |
| use self-attention Fusion Blocks instead of symmetric cross attention (ablation) | AV-Link V2A · VGGSounds V2A ablation | FAD: 8.47 -> 4.79 | Improved |
| use Fusion Blocks instead of concatenation to text with fine-tuning (ablation) | AV-Link V2A · VGGSounds V2A ablation | Onset ACC: 0.355 -> 0.415 | Improved |
| interleave Fusion Blocks instead of placing them after block 22 (ablation) | AV-Link V2A · VGGSounds V2A ablation | FAD: 6.90 -> 4.79 | Improved |
| train separate parameters for V2A and A2V instead of joint sharing (ablation) | AV-Link V2A · VGGSounds V2A (no text prompt) | FAD: 2.19 -> 1.58 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Frieren — Frieren: Efficient Video-to-Audio Generation Network with Rectified Flow Matching (NeurIPS 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Seeing and Hearing: Open-domain Visual-Audio Generation with Diffusion Latent Aligners (CVPR 2024 · similarity 0.90)
- MMDisCo: Multi-Modal Discriminator-Guided Cooperative Diffusion for Joint Audio and Video Generation (ICLR 2025 · similarity 0.90)
- Foley-Flow: Coordinated Video-to-Audio Generation with Masked Audio-Visual Alignment and Dynamic Conditional Flows (CVPR 2025 · similarity 0.88)
- Native Audio-Visual Alignment for Generation (NeurIPS 2026 · similarity 0.88)
- Diverse and Aligned Audio-to-Video Generation via Text-to-Video Model Adaptation (AAAI 2024 · similarity 0.87)
- Asymmetric Hierarchical Anchoring for Robust Audio–Visual Cross-Modal Generalization (NeurIPS 2026 · similarity 0.87)
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.
