Research map / 3D vision / 3D generation
Learning Neural Acoustic Fields
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 |
|---|---|---|---|
| share one local feature grid between emitter and listener instead of using separate grids (ablation) | NAF · Soundspaces, mean | spectral loss: 0.388 -> 0.380 | Helped |
| share one local feature grid between emitter and listener instead of using separate grids (ablation) | NAF · Soundspaces, mean | T60 error: 4.241 -> 3.650 | Helped |
| add local geometric conditioning (ablation) | NAF · Soundspaces, Large 1 | spectral loss | Helped |
| predict the impulse response in the STFT domain instead of directly in the time domain (ablation) | NAF · Soundspaces, Large 1 | spectral loss: 2.046 -> 0.396 | Helped |
| predict the impulse response in the STFT domain instead of directly in the time domain (ablation) | NAF · Soundspaces, Large 1 | T60 error: 49.72 -> 4.166 | Helped |
| add L2 regularization to the NeRF feature grid (ablation) | NeRF + grid · Large Room 1, 75 training images | PSNR: 25.41 -> 22.69 | Hurt |
| add L2 regularization to the NeRF feature grid (ablation) | NeRF + grid · Large Room 2, 75 training images | PSNR: 25.70 -> 24.86 | Hurt |
| jointly train NeRF with acoustic supervision from NAF | NeRF · Large Room 1, 75 training images | PSNR: 25.41 -> 26.19 | Helped |
| jointly train NeRF with acoustic supervision from NAF | NeRF · Large Room 2, 75 training images | PSNR: 25.70 -> 26.24 | Helped |
| use NAF latents instead of MFCC features for linear decoding | scene structure decoder · Soundspaces, mean | explained variance: 0.626 -> 0.913 | Helped |
| represent the acoustic field with NAF instead of Opus coding | Soundspaces · Soundspaces, mean | storage (MiB): 181.37 -> 8.41 | Helped |
Closest papers by meaning
- Real Acoustic Fields: An Audio-Visual Room Acoustics Dataset and Benchmark (CVPR 2024 · similarity 0.89)
- NeRAF: 3D Scene Infused Neural Radiance and Acoustic Fields (ICLR 2025 · similarity 0.88)
- Resounding Acoustic Fields with Reciprocity (NeurIPS 2025 · similarity 0.87)
- Deep Neural Room Acoustics Primitive (ICML 2024 · similarity 0.86)
- Point Clouds Meets Physics: Dynamic Acoustic Field Fitting Network for Point Cloud Understanding (CVPR 2025 · similarity 0.86)
- SoundVista: Novel-View Ambient Sound Synthesis via Visual-Acoustic Binding (CVPR 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.