Research map / Vision-language and multimodal / Cross-modal learning
SAM Audio: Segment Anything in Audio
Bowen Shi, Andros Tjandra, John Hoffman, Helin Wang, YI-CHIAO WU, Luya Gao, Julius Richter, Matthew Le, Apoorv Vyas, Sanyuan Chen, Christoph Feichtenhofer, Piotr Dollár and 2 more
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 ground-truth span to text prompt (ablation) | SAM Audio · General sound | OVR subjective score: 3.32 -> 4.04 | Improved |
| use span prompt alone instead of text on music (ablation) | SAM Audio · Music | OVR subjective score: 4.09 -> 2.57 | Hurt |
| use span prompt alone instead of text on general sound (ablation) | SAM Audio · General sound | OVR subjective score: 3.32 -> 3.54 | Improved |
| add PEA-Frame predicted spans to text prompting (ablation) | SAM Audio · General SFX | OVR subjective score: 3.36 -> 3.89 | Improved |
| add PEA-Frame predicted spans to text prompting (ablation) | SAM Audio · Speaker | OVR subjective score: 3.62 -> 4.01 | Improved |
| add PEA-Frame predicted spans to text prompting (ablation) | SAM Audio · Instrument (pro, MUSDB) | OVR subjective score: 4.16 -> 4.12 | Hurt |
| add predicted spans to text prompting (ablation) | SAM Audio · Speech | OVR subjective score: 4.17 -> 4.22 | No clear effect |
| use predicted spans instead of ground-truth spans (ablation) | SAM Audio · text+span prompting | OVR subjective score | No clear effect |
| use visual prompt instead of text prompt | SAM Audio · General SFX | OVR subjective score: 3.59 (text) -> 2.61 (visual) | Hurt |
| SAM Audio over specialized models | text-prompted separation · Instrument (pro, MUSDB) | OVR subjective score: 4.28 -> 4.45 | Improved |
| use SAM Audio Judge instead of CLAP/SDR estimator/Gemini as metric | reference-free evaluation · Speech | Pearson correlation with humans: 0.490 -> 0.883 | 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.
- CLAP — CLAP: Isolating Content from Style through Contrastive Learning with Augmented Prompts (ECCV 2024) · better on 1 of 1 shared result
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
- Massive Sound Embedding Benchmark (MSEB) (NeurIPS 2025 · similarity 0.88)
- Listen, Think, and Understand (ICLR 2024 · similarity 0.88)
- Weakly-supervised Audio Separation via Bi-modal Semantic Similarity (ICLR 2024 · similarity 0.88)
- MMAU: A Massive Multi-Task Audio Understanding and Reasoning Benchmark (ICLR 2025 · similarity 0.88)
- Continual Audio-Visual Sound Separation (NeurIPS 2024 · similarity 0.88)
- Unified Synthesis of Compositional Speech and Sound from Free-Form Text Prompts (NeurIPS 2026 · similarity 0.88)
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.
