Urban Sound Propagation

Urban Sound Propagation数据集专注于城市环境中声音传播的复杂模式和行为，由机器学习与分析研究所(IMLA)创建。该数据集包含100,000个样本，每个样本包括来自OpenStreetMap的真实2D建筑地图、参数化声源和模拟的地面真实声音传播。数据集通过四个不同的模拟任务，逐步增加反射、衍射和源变异性的复杂性。该数据集的创建过程涉及从OpenStreetMap下载数据，使用NoiseModelling v4.0框架进行声音传播模拟，并最终生成声音分布图。该数据集的应用领域包括城市规划、噪声污染管理和声音映射技术的改进，旨在解决城市环境中声音传播的复杂物理问题。The Urban Sound Propagation dataset, created by the Institute of Machine Learning and Analytics (IMLA), focuses on the complex patterns and behaviors of sound propagation in urban environments. It contains 100,000 samples, each of which includes real 2D building maps sourced from OpenStreetMap, parametric sound sources, and simulated ground-truth sound propagation. The dataset gradually escalates the complexity of reflections, diffraction, and source variability through four distinct simulation tasks. The creation process of this dataset involves downloading data from OpenStreetMap, performing sound propagation simulations using the NoiseModelling v4.0 framework, and finally generating sound distribution maps. This dataset has applications in urban planning, noise pollution management, and the advancement of sound mapping technologies, aiming to resolve the complex physical problems related to sound propagation in urban environments.

搜集汇总

背景与挑战

背景概述

Urban Sound Propagation数据集是一个用于评估一步生成学习模型在城市声音传播建模中的基准数据集。它包含25,000个数据点，覆盖10个城市，每个点提供基于OpenStreetMap的建筑布局图像和通过NoiseModelling v4.0模拟生成的声音分布图，支持反射、衍射和声源变化等复杂物理系统的研究。