Research map / Video / Video generation
Video Echoed in Music: Semantic, Temporal, and Rhythmic Alignment for Video-to-Music Generation
Xinyi Tong, Yiran Zhu, Jishang Chen, Chunru Zhan, Tianle Wang, Sirui Zhang, Nian Liu, Tiezheng Ge, Duo XU, Xin Jin, Feng Yu, Song-Chun Zhu
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 |
|---|---|---|---|
| add transition-beat aligner and adapter (TB-As) (ablation) | VeM · TB-Match validation | TBIoU: 0.265 -> 0.364 | Improved |
| add TB-As to unconditional model (ablation) | VeM without HVP-Cond and SG-CAtt · TB-Match validation | BIoU: 0.221 -> 0.433 | Improved |
| add TB-As to unconditional model (ablation) | VeM without HVP-Cond and SG-CAtt · TB-Match validation | FAD: 6.692 -> 7.217 | Hurt |
| add hierarchical video parsing conditions and SG-CAtt (ablation) | VeM without TB-As · TB-Match validation | FAD: 6.692 -> 4.382 | Improved |
| replace standard cross-attention with SG-CAtt (ablation) | VeM with HVP-Cond · TB-Match validation | LB: 0.735 -> 0.890 | Improved |
| replace standard cross-attention with SG-CAtt (ablation) | VeM with HVP-Cond · TB-Match validation | FAD: 5.712 -> 4.382 | Improved |
| add TB-As on top of HVP-Cond and SG-CAtt (ablation) | VeM · TB-Match validation | BIoU: 0.403 -> 0.594 | Improved |
| add TB-As on top of HVP-Cond and SG-CAtt (ablation) | VeM · TB-Match validation | FAD: 4.382 -> 4.043 | Improved |
| add hierarchical video parsing conditions alone (ablation) | VeM without SG-CAtt and TB-As · TB-Match validation | CLAP: 0.180 -> 0.218 | Improved |
| compare VeM against GVMGen on rhythmic alignment | VeM · TB-Match validation | TBIoU: 0.357 -> 0.364 | Improved |
| compare VeM against GVMGen on beat IoU | VeM · TB-Match validation | BIoU: 0.465 -> 0.594 | Improved |
| evaluate zero-shot on Sora and SymMV videos | VeM · SymMV, Sora, other videos | LB and TBIoU: SymMV LB 0.989 vs 0.925 (M2UGen) | 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.
- VidMuse — VidMuse: A Simple Video-to-Music Generation Framework with Long-Short-Term Modeling (CVPR 2025) · better on 5 of 5 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
- V2Meow: Meowing to the Visual Beat via Video-to-Music Generation (AAAI 2024 · similarity 0.90)
- VidMuse: A Simple Video-to-Music Generation Framework with Long-Short-Term Modeling (CVPR 2025 · similarity 0.89)
- HarmonySet: A Comprehensive Dataset for Understanding Video-Music Semantic Alignment and Temporal Synchronization (CVPR 2025 · similarity 0.89)
- GVMGen: A General Video-to-Music Generation Model with Hierarchical Attentions (AAAI 2025 · similarity 0.87)
- MoMu-Diffusion: On Learning Long-Term Motion-Music Synchronization and Correspondence (NeurIPS 2024 · similarity 0.86)
- Customized Condition Controllable Generation for Video Soundtrack (CVPR 2025 · similarity 0.86)
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.
