Research map / Vision-language and multimodal / Cross-modal learning
Recursive Visual Sound Separation Using Minus-Plus Net
Xudong Xu, Bo Dai, Dahua Lin
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 P-Net (compensating for removed shared sound patterns) to M-Net (ablation) | MP-Net (binary mask) · VEGAS (mix-2) | AMID (lower better): 5.96 -> 5.78 | Improved |
| add P-Net (compensating for removed shared sound patterns) to M-Net (ablation) | MP-Net (binary mask) · VEGAS (mix-2) | NSDR: 2.02 -> 2.14 | Improved |
| add P-Net (compensating for removed shared sound patterns) to M-Net (ablation) | MP-Net (binary mask) · VEGAS (mix-3) | NSDR: 1.23 -> 1.48 | Improved |
| add P-Net (compensating for removed shared sound patterns) to M-Net (ablation) | MP-Net (binary mask) · MUSIC (mix-2) | NSDR: 5.47 -> 5.73 | Improved |
| add P-Net (compensating for removed shared sound patterns) to M-Net (ablation) | MP-Net (binary mask) · MUSIC (mix-3) | SIR: 7.89 -> 8.18 | Improved |
| switch from binary masks to ratio masks | MP-Net · VEGAS (mix-2) | NSDR / SIR: NSDR 2.14 -> 2.81; SIR 7.66 -> 5.45 | Mixed |
| recursive energy-ordered separation (MP-Net) instead of independent separation (PixelPlayer) | visual sound separation · MUSIC (mix-2) | NSDR (binary): 5.16 -> 5.73 | Improved |
| recursive energy-ordered separation (MP-Net) instead of independent separation (PixelPlayer) | visual sound separation · VEGAS (mix-3) | SIR (binary): 4.82 -> 4.99 | Improved |
| evaluate with more sounds in the mixture (up to 5) than trained on | MP-Net · MUSIC and VEGAS | separation scores | Improved |
Closest papers by meaning
- High-Quality Sound Separation Across Diverse Categories via Visually-Guided Generative Modeling (IJCV 2026 · similarity 0.87)
- Into the Wild with AudioScope: Unsupervised Audio-Visual Separation of On-Screen Sounds (ICLR 2021 · similarity 0.87)
- RTFS-Net: Recurrent Time-Frequency Modelling for Efficient Audio-Visual Speech Separation (ICLR 2024 · similarity 0.86)
- Continual Audio-Visual Sound Separation (NeurIPS 2024 · similarity 0.85)
- Dynamic Derivation and Elimination: Audio Visual Segmentation with Enhanced Audio Semantics (CVPR 2025 · similarity 0.84)
- Independency Adversarial Learning for Cross-Modal Sound Separation (AAAI 2024 · similarity 0.84)
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.
