Real-time Prediction of Urban Sound Propagation with Conditioned Normalizing Flows

2025-10-06

AI Summary

To address the challenges of real-time urban noise prediction and low accuracy in non-line-of-sight (NLOS) regions, this paper proposes the first conditional normalizing flow method for acoustic propagation modeling—built upon the Full-Glow architecture—that takes 2D urban layouts as conditional inputs and generates high-fidelity sound pressure distribution maps end-to-end. The model achieves an inference speed of 256×256 sound pressure maps per second on a single RTX 4090 GPU, accelerating over traditional numerical simulations by more than 2000×. In NLOS scenarios, it attains a mean absolute error (MAE) of 0.65 dB—up to 24% improvement over state-of-the-art deep learning methods—accurately capturing diffraction and interference effects while complying with the EU Environmental Noise Directive mapping requirements. Crucially, the method supports instantaneous recomputation under dynamic source configurations and geometric changes, enabling differentiable, interactive “what-if” analysis for urban planning and noise mitigation.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Sampling/Simulation-based Search
• Natural Language Processing: Generation

Application Category

• 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
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Accurate and fast urban noise prediction is pivotal for public health and for regulatory workflows in cities, where the Environmental Noise Directive mandates regular strategic noise maps and action plans, often needed in permission workflows, right-of-way allocation, and construction scheduling. Physics-based solvers are too slow for such time-critical, iterative "what-if" studies. We evaluate conditional Normalizing Flows (Full-Glow) for generating for generating standards-compliant urban sound-pressure maps from 2D urban layouts in real time per 256x256 map on a single RTX 4090), enabling interactive exploration directly on commodity hardware. On datasets covering Baseline, Diffraction, and Reflection regimes, our model accelerates map generation by >2000 times over a reference solver while improving NLoS accuracy by up to 24% versus prior deep models; in Baseline NLoS we reach 0.65 dB MAE with high structural fidelity. The model reproduces diffraction and interference patterns and supports instant recomputation under source or geometry changes, making it a practical engine for urban planning, compliance mapping, and operations (e.g., temporary road closures, night-work variance assessments).

Problem

• Predicting urban noise propagation in real time
• Accelerating noise map generation for urban planning
• Enabling interactive sound simulations on consumer hardware

Innovation

• Uses conditional Normalizing Flows for sound prediction
• Generates real-time sound maps from 2D urban layouts
• Accelerates computation by 2000x while improving accuracy