Research map / Video / Video segmentation and tracking
Co-Separating Sounds of Visual Objects
Ruohan Gao, Kristen Grauman
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 duet (multi-source) videos to solo training data | Co-Separation · MUSIC test | SDR: 7.38 -> 7.64 | Improved |
| add duet (multi-source) videos to solo training data | Sound-of-Pixels · MUSIC test | SDR: 7.30 -> 6.05 | Hurt |
| add duet videos to solo training data | AV-Mix-and-Separate · MUSIC test | SDR: 3.16 -> 3.23 | No clear effect |
| use co-separation training instead of video-level mix-and-separate | same separator network · MUSIC test (solo + duet) | SDR: 3.23 -> 7.64 | Improved |
| use co-separation training instead of Sound-of-Pixels | Sound-of-Pixels baseline · AudioSet-SingleSource | SDR: 1.66 -> 4.26 | Improved |
| use co-separation on noisy AudioSet instead of AV-Mix-and-Separate | AV-Mix-and-Separate · AudioSet-SingleSource | SDR: 1.68 -> 4.26 | Improved |
| co-separation training for objects never heard individually (Violin/Saxophone) | Sound-of-Pixels · Unseen-solo toy test | SDR: 1.52 -> 8.10 | Improved |
| co-separation training for Violin/Guitar mixture | Sound-of-Pixels · Unseen-solo toy test | SDR: 6.95 -> 10.6 | Improved |
| co-separation training for Saxophone/Guitar mixture | Sound-of-Pixels · Unseen-solo toy test | SDR: 0.57 -> 5.08 | Improved |
| co-separation versus artifact-free baselines on SAR | Co-Separation · AudioSet / MUSIC | SAR: SAR 13.0 vs 12.2 on AudioSet; 10.8 vs 11.9 on MUSIC solo | Mixed |
| co-separation versus AV-MIML on denoising | Co-Separation · AV-Bench Wooden Horse | NSDR: 12.3 -> 14.5 | Improved |
| co-separation versus AV-Loc on Guitar Solo denoising | Co-Separation · AV-Bench Guitar Solo | NSDR: 14.1 vs 11.9 | Hurt |
Closest papers by meaning
- Learning to Separate Object Sounds by Watching Unlabeled Video (ECCV 2018 · similarity 0.94)
- Into the Wild with AudioScope: Unsupervised Audio-Visual Separation of On-Screen Sounds (ICLR 2021 · similarity 0.91)
- Seeing Speech and Sound: Distinguishing and Locating Audio Sources in Visual Scenes (CVPR 2025 · similarity 0.87)
- DeepASA: An Object-Oriented Multi-Purpose Network for Auditory Scene Analysis (NeurIPS 2025 · similarity 0.87)
- Weakly-supervised Audio Separation via Bi-modal Semantic Similarity (ICLR 2024 · similarity 0.86)
- Continual Audio-Visual Sound Separation (NeurIPS 2024 · 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.
