Research map / Video / Video generation
Hear What Matters! Text-conditioned Selective Video-to-Audio Generation
Junwon Lee, Juhan Nam, Jiyoung Lee
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 video encoder finetuning (text-conditioned) (ablation) | SELVA · SelVA-Bench (inter-class) | DeSync: 0.868 -> 0.721 | Improved |
| add video encoder finetuning (text-conditioned) (ablation) | SELVA · SelVA-Bench (intra-class) | DeSync: 0.734 -> 0.639 | Improved |
| add V2A generator finetuning with auto-mixed samples (ablation) | SELVA · SelVA-Bench (inter-class) | FAD: 56.6 -> 51.7 | Improved |
| add [SUP] supplementary tokens (ablation) | SELVA · SelVA-Bench (inter-class) | DeSync: 0.756 -> 0.721 | Improved |
| add [SUP] supplementary tokens (ablation) | SELVA · SelVA-Bench (intra-class) | DeSync: 0.676 -> 0.639 | Improved |
| add [SUP] supplementary tokens (ablation) | SELVA · SelVA-Bench (inter-class) | KAD: 0.637 -> 0.676 | Hurt |
| use two-stage training instead of joint training (ablation) | SELVA · SelVA-Bench (inter-class) | DeSync: 0.823 -> 0.721 | Improved |
| use two-stage training instead of joint training (ablation) | SELVA · SelVA-Bench (intra-class) | DeSync: 0.777 -> 0.639 | Improved |
| increase [SUP] tokens from 0/1/3/7 to 5 (ablation) | SELVA · SelVA-Bench (inter-class) | DeSync: 0.756 -> 0.721 | Improved |
| replace 5 tokens by 7 tokens (ablation) | SELVA · SelVA-Bench (intra-class) | DeSync: 0.639 -> 0.675 | Hurt |
| remove CLIP embedding (null) from MMAudio (ablation) | MMAudio-S-16kHz · VGGSound test | CLAP: 0.197 -> 0.235 | Improved |
| remove Synchformer embedding (null) from MMAudio (ablation) | MMAudio-S-16kHz · VGGSound test | DeSync: 0.492 -> 1.243 | Hurt |
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.
- VinTAGe — VinTAGe: Joint Video and Text Conditioning for Holistic Audio Generation (CVPR 2025) · 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
- Diverse and Aligned Audio-to-Video Generation via Text-to-Video Model Adaptation (AAAI 2024 · similarity 0.88)
- VinTAGe: Joint Video and Text Conditioning for Holistic Audio Generation (CVPR 2025 · similarity 0.87)
- Read, Watch and Scream! Sound Generation from Text and Video (AAAI 2025 · similarity 0.87)
- Tell What You Hear From What You See - Video to Audio Generation Through Text (NeurIPS 2024 · similarity 0.87)
- MMAudio: Taming Multimodal Joint Training for High-Quality Video-to-Audio Synthesis (CVPR 2025 · similarity 0.87)
- Audio-Sync Video Generation with Multi-Stream Temporal Control (NeurIPS 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.
