SINGA:PURA

SINGA:PURA是一个包含时空上下文的强标注多音城市声音数据集，由新加坡南洋理工大学电气与电子工程学院创建。该数据集包含6547条在真实声学环境中录制的音频，每条音频附带时空元数据。数据通过部署在新加坡的无线声学传感器网络收集，旨在识别和缓解噪音源，同时也适用于声音事件检测、分类和定位。数据集的创建涉及详细的数据收集、标注和处理方法，适用于城市声音标记系统的训练和评估，特别关注解决城市环境中复杂声音事件的问题。SINGA:PURA is a densely annotated multi-sound urban sound dataset with spatio-temporal context, developed by the School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore. This dataset comprises 6,547 audio recordings captured in real-world acoustic environments, with each recording paired with corresponding spatio-temporal metadata. The data was collected through a wireless acoustic sensor network deployed across Singapore, which is intended to identify and mitigate noise sources, while also being applicable to tasks including sound event detection, classification and localization. The creation of this dataset entails rigorous data collection, annotation and processing workflows, and serves as a benchmark for training and evaluating urban sound tagging systems, with a special focus on tackling the challenges posed by complex sound events in urban settings.

搜集汇总

背景与挑战

背景概述

SINGA:PURA是一个强标签的多音城市声音数据集，具有时空上下文，数据通过部署在新加坡的无线声学传感器网络收集，旨在支持噪声源识别和声音事件检测等应用。其标签分类兼容现有标准，同时包含新加坡特有的声音事件，适用于工程和物理领域的研究。