Research map / Video / Video generation
Video Background Music Generation: Dataset, Method and Evaluation
Le Zhuo, Zhaokai Wang, Baisen Wang, Yue Liao, Chenxi Bao, Stanley Peng, Songhao Han, Aixi Zhang, Fei Fang, Si Liu
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 |
|---|---|---|---|
| decouple music generation into chord, melody and accompaniment stages instead of direct video-to-music (ablation) | V-MusProd · SymMV test set | VMCP P@20: 33.3 -> 44.8 | Improved |
| decouple chords before music (Video2chord2music) versus direct generation (ablation) | V-MusProd · SymMV test set | VMCP P@5: 10.8 -> 13.7 | Improved |
| decouple melody and accompaniment in addition to chords (ablation) | V-MusProd · SymMV test set | VMCP P@5: 13.7 -> 15.7 | Improved |
| add semantic features to video controller (ablation) | V-MusProd · SymMV test set | VMCP P@5: 11.6 -> 15.7 | Improved |
| add motion features (tempo and timing encoding) (ablation) | V-MusProd · SymMV test set | VMCP P@5: 12.2 -> 15.7 | Improved |
| add color features to video controller (ablation) | V-MusProd · SymMV test set | VMCP P@5: 15.6 -> 15.7 | No clear effect |
| add decoupled structure and video control versus direct generation on music quality (ablation) | Video2music · SymMV test set | Pitch Entropy: 3.990 (Video2music) vs 3.940 (V-MusProd), real 4.197 | Mixed |
| use V-MusProd instead of CMT for video-conditioned music | CMT baseline · SymMV test set | VMCP P@5: 8.9 -> 15.7 | Improved |
| use V-MusProd decoupled structure for unconditional generation versus CP Transformer | unconditional music generation · POP909 unconditional generation | Empty Beat Rate: 0.041 -> 0.005 | Improved |
Closest papers by meaning
- GVMGen: A General Video-to-Music Generation Model with Hierarchical Attentions (AAAI 2025 · similarity 0.91)
- VidMuse: A Simple Video-to-Music Generation Framework with Long-Short-Term Modeling (CVPR 2025 · similarity 0.90)
- V2Meow: Meowing to the Visual Beat via Video-to-Music Generation (AAAI 2024 · similarity 0.90)
- HarmonySet: A Comprehensive Dataset for Understanding Video-Music Semantic Alignment and Temporal Synchronization (CVPR 2025 · similarity 0.87)
- Customized Condition Controllable Generation for Video Soundtrack (CVPR 2025 · similarity 0.87)
- FilmComposer: LLM-Driven Music Production for Silent Film Clips (CVPR 2025 · 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.
