Differentiable Acoustic Radiance Transfer

2025-09-19

AI Summary

This work addresses the challenge of jointly optimizing material properties in room acoustics modeling. We propose Differentiable Acoustic Radiative Transfer (DART), a physics-informed method grounded in geometric acoustics and radiative transfer theory. DART establishes a time-varying, directionally resolved energy interaction model among discretized surface patches and—crucially—enables fully differentiable end-to-end solution of the acoustic rendering equation for the first time. The method supports gradient-based optimization of wall-mounted material absorption and scattering coefficients directly from sparse acoustic measurements, enabling high-fidelity prediction of energy responses for arbitrary novel source–receiver configurations. Compared to conventional signal-processing approaches and black-box neural networks, DART achieves superior generalization, explicit physical interpretability, and computational efficiency. It establishes a new paradigm for data-driven, differentiable acoustic modeling.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Geometric acoustics is an efficient approach to room acoustics modeling, governed by the canonical time-dependent rendering equation. Acoustic radiance transfer (ART) solves the equation through discretization, modeling the time- and direction-dependent energy exchange between surface patches given with flexible material properties. We introduce DART, a differentiable and efficient implementation of ART that enables gradient-based optimization of material properties. We evaluate DART on a simpler variant of the acoustic field learning task, which aims to predict the energy responses of novel source-receiver settings. Experimental results show that DART exhibits favorable properties, e.g., better generalization under a sparse measurement scenario, compared to existing signal processing and neural network baselines, while remaining a simple, fully interpretable system.

Problem

• Differentiable acoustic radiance transfer for material optimization
• Solving energy exchange in room acoustics modeling
• Predicting acoustic responses in novel source-receiver settings

Innovation

• Differentiable Acoustic Radiance Transfer implementation
• Gradient-based optimization of material properties
• Interpretable system for acoustic field learning