AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

p-AVAS: Can Physics-Integrated Audio-Visual Modeling Boost Neural Acoustic Synthesis?

Susan Liang, Chao Huang, Yunlong Tang, Zeliang Zhang, Chenliang Xu

ICCV 2025 · Vision-language and multimodal › Cross-modal learning · approach: Audio-visual source localization

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 changedWhereMeasuredEffect
add vision-guided audio simulation as inputINRAS · RWAVS-Gen (overall)MAG: 3.081 -> 2.763Improved
add vision-guided audio simulation as inputNAF · RWAVS-Gen (overall)MAG: 2.900 -> 2.724Improved
add vision-guided audio simulation as inputViGAS · RWAVS-Gen (overall)MAG: 2.982 -> 2.438Improved
add vision-guided audio simulation as inputAV-NeRF · RWAVS-Gen (overall)MAG: 3.168 -> 2.396Improved
add vision-guided audio simulation as inputAV-NeRF · RWAVS-Gen (overall)ENV: 0.199 -> 0.163Improved
use mesh-based simulation instead of pyroomacoustics shoebox + HRTF (ablation)pi-AVAS · RWAVS-Gen (overall), ablationMAG: 5.435 -> 3.938Improved
use flow matching loss instead of regression loss for refinement (ablation)pi-AVAS refinement network · RWAVS-Gen (overall), ablationMAG: 2.844 -> 2.702Improved
add flow matching refinement to simulation (ablation)pi-AVAS · RWAVS-Gen (overall), ablationMAG: 3.938 -> 2.360Improved
add simulation stage to flow matching refinement (ablation)pi-AVAS · RWAVS-Gen (overall), ablationMAG: 2.702 -> 2.360Improved
add data augmentation (ablation)pi-AVAS · RWAVS-Gen (overall), ablationMAG: 2.360 -> 2.336Improved
add data augmentation on office and apartment scenes (ablation)pi-AVAS · RWAVS-Gen office/apartmentMAGNo clear effect
add data augmentation on house and outdoor scenes (ablation)pi-AVAS · RWAVS-Gen house/outdoorMAG: 3.144 -> 1.946 (house)Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy