Research map / Detection and segmentation / Open-set and few-shot detection
RealImpact: A Dataset of Impact Sound Fields for Real Objects
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use explicit acoustic-transfer modeling rather than modeling structural vibration alone | Object impact sound simulation · RealImpact Deconvolved and Deconvolved + Denoised | L1 Spectral: KleinPat 0.747 -> 0.632; NeuralSound 0.747 -> 0.673; denoised: KleinPat 0.236 -> 0.0982; NeuralSound 0.236 -> 0.133 | Helped |
| Use explicit acoustic-transfer modeling rather than modeling structural vibration alone | Object impact sound simulation · RealImpact Deconvolved and Deconvolved + Denoised | CDPAM: KleinPat 0.297 -> 0.117; NeuralSound 0.297 -> 0.102; denoised: KleinPat 0.289 -> 0.0975; NeuralSound 0.289 -> 0.0750 | Helped |
| Compare against denoised rather than raw deconvolved recordings | Simulation baseline evaluation · RealImpact recordings | L1 Spectral: Random Impact Sound 0.728 -> 0.150; KleinPat 0.632 -> 0.0982; NeuralSound 0.673 -> 0.133; ObjectFolder 2.0 0.747 -> 0.23 | Helped |
| Compare against denoised rather than raw deconvolved recordings | Simulation baseline evaluation · RealImpact recordings | Envelope (× 10^-3): Random Impact Sound 4.17 -> 4.97; KleinPat 4.63 -> 4.63; NeuralSound 23.0 -> 22.8; ObjectFolder 2.0 25.6 -> 25.4 | Mixed |
| Compare against denoised rather than raw deconvolved recordings | Simulation baseline evaluation · RealImpact recordings | CDPAM: Random Impact Sound 0.121 -> 0.0880; KleinPat 0.117 -> 0.0975; NeuralSound 0.102 -> 0.0750; ObjectFolder 2.0 0.297 -> 0. | Helped |
| Use a random impact recording instead of a simulation | Object impact sound comparison · RealImpact Deconvolved | Envelope (× 10^-3): 4.63 -> 4.17 | Helped |
| Use a random impact recording instead of a simulation | Object impact sound comparison · RealImpact Deconvolved + Denoised | Envelope (× 10^-3): 4.63 -> 4.97 | Hurt |
| Predict listener position from the impact sound | RealImpact listener location classification · Angle, height, and distance | Accuracy (%): 10.0 -> 57.9; 6.7 -> 60.7; 25.0 -> 67.4 | Helped |
| Match impact recordings to corresponding RGB images | RealImpact visual acoustic matching · Visual Acoustic Matching | Accuracy (%): 50.0 -> 75.1 | Helped |
| Match impact recordings to corresponding RGB images | RealImpact visual acoustic matching · Visual Acoustic Matching | RMSE angle error (degrees): 59.7 -> 47.7 | Helped |
Closest papers by meaning
- Real Acoustic Fields: An Audio-Visual Room Acoustics Dataset and Benchmark (CVPR 2024 · similarity 0.88)
- SonicSim: A customizable simulation platform for speech processing in moving sound source scenarios (ICLR 2025 · similarity 0.85)
- SoundVista: Novel-View Ambient Sound Synthesis via Visual-Acoustic Binding (CVPR 2025 · similarity 0.84)
- Towards Audio-Visual Navigation in Noisy Environments: A Large-Scale Benchmark Dataset and an Architecture Considering Multiple Sound-Sources (AAAI 2025 · similarity 0.83)
- Hearing Anywhere in Any Environment (CVPR 2025 · similarity 0.83)
- Hearing Hands: Generating Sounds from Physical Interactions in 3D Scenes (CVPR 2025 · similarity 0.83)
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.