Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Urban Sound Propagation: a Benchmark for 1-Step Generative Modeling of Complex Physical Systems

Martin Spitznagel, Janis Keuper

TL;DR

This work presents a novel benchmark for the evaluation of 1-step generative learning models in terms of speed and physical correctness, and shows that while these models are very well capable of modeling sound propagations in simple cases, the approximation of sub-systems represented by higher order equations systematically fails.

Abstract

Data-driven modeling of complex physical systems is receiving a growing amount of attention in the simulation and machine learning communities. Since most physical simulations are based on compute-intensive, iterative implementations of differential equation systems, a (partial) replacement with learned, 1-step inference models has the potential for significant speedups in a wide range of application areas. In this context, we present a novel benchmark for the evaluation of 1-step generative learning models in terms of speed and physical correctness. Our Urban Sound Propagation benchmark is based on the physically complex and practically relevant, yet intuitively easy to grasp task of modeling the 2d propagation of waves from a sound source in an urban environment. We provide a dataset with 100k samples, where each sample consists of pairs of real 2d building maps drawn from OpenStreetmap, a parameterized sound source, and a simulated ground truth sound propagation for the given scene. The dataset provides four different simulation tasks with increasing complexity regarding reflection, diffraction and source variance. A first baseline evaluation of common generative U-Net, GAN and Diffusion models shows, that while these models are very well capable of modeling sound propagations in simple cases, the approximation of sub-systems represented by higher order equations systematically fails. Information about the dataset, download instructions and source codes are provided on our website: https://www.urban-sound-data.org.

Urban Sound Propagation: a Benchmark for 1-Step Generative Modeling of Complex Physical Systems

TL;DR

This work presents a novel benchmark for the evaluation of 1-step generative learning models in terms of speed and physical correctness, and shows that while these models are very well capable of modeling sound propagations in simple cases, the approximation of sub-systems represented by higher order equations systematically fails.

Abstract

Data-driven modeling of complex physical systems is receiving a growing amount of attention in the simulation and machine learning communities. Since most physical simulations are based on compute-intensive, iterative implementations of differential equation systems, a (partial) replacement with learned, 1-step inference models has the potential for significant speedups in a wide range of application areas. In this context, we present a novel benchmark for the evaluation of 1-step generative learning models in terms of speed and physical correctness. Our Urban Sound Propagation benchmark is based on the physically complex and practically relevant, yet intuitively easy to grasp task of modeling the 2d propagation of waves from a sound source in an urban environment. We provide a dataset with 100k samples, where each sample consists of pairs of real 2d building maps drawn from OpenStreetmap, a parameterized sound source, and a simulated ground truth sound propagation for the given scene. The dataset provides four different simulation tasks with increasing complexity regarding reflection, diffraction and source variance. A first baseline evaluation of common generative U-Net, GAN and Diffusion models shows, that while these models are very well capable of modeling sound propagations in simple cases, the approximation of sub-systems represented by higher order equations systematically fails. Information about the dataset, download instructions and source codes are provided on our website: https://www.urban-sound-data.org.
Paper Structure (14 sections, 2 equations, 10 figures, 6 tables)

This paper contains 14 sections, 2 equations, 10 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Physics of Sound Propagation
  4. Dataset Creation and Properties
  5. Prediction Tasks
  6. Evaluation & Baseline Results
  7. Experimental Setup
  8. Generative Models.
  9. Baseline Performance
  10. Analysis
  11. Discussion & Future Work
  12. Scalable Simulation Pipeline.
  13. Additional Qualitative Results
  14. Training Setup

Figures (10)

  • Figure 1: Starting with the selection of a 500m² area (a), buildings are identified, followed by placing a receiver grid (b). The urban layout (c) and the corresponding sound propagation, simulated using the NoiseModelling Framework from a central signal source, are then used in the dataset (d).
  • Figure 2: Comparing the ground-truth simulation with the predictions from U-Net, GAN, and diffusion model for a single sample within the reflection task.
  • Figure 3: Comparing the ground-truth simulation with the prediction of the diffusion model for a single sample within the combined task, distinguishing between the MAE in LoS and NLoS.
  • Figure 4: Comparing the ground-truth simulation with the prediction of the diffusion model for a single sample within the reflection task, distinguishing between the MAE in LoS and NLoS.
  • Figure 5: Detailed visualization of the dataset generation pipeline.
  • ...and 5 more figures