An Open-Set Recognition and Few-Shot Learning Dataset for Audio Event Classification in Domestic Environments

本数据集名为‘An Open-Set Recognition and Few-Shot Learning Dataset for Audio Event Classification in Domestic Environments’，由瓦伦西亚大学创建，包含1360个音频片段，分为34个类别，主要用于家庭环境中的音频事件分类研究。数据集涵盖了多种音频警报，如门铃或火灾警报，旨在通过有限样本进行特定和有意音频事件的检测。此外，数据集还支持开放集识别（OSR）算法的评估，允许在实际应用中预测训练阶段未见过的类别的样本。

This dataset is titled "An Open-Set Recognition and Few-Shot Learning Dataset for Audio Event Classification in Domestic Environments", and was created by the University of Valencia. It comprises 1360 audio clips spanning 34 categories, and is primarily used for research on audio event classification in domestic environments. The dataset covers various audio alarms such as doorbells and fire alarms, and is designed to detect specific intentional audio events using limited samples. Additionally, it supports the evaluation of open-set recognition (OSR) algorithms, enabling the prediction of samples belonging to categories unseen during the training phase in real-world applications.