WE-LIVE
收藏资源简介:
WE-LIVE数据集是由马德里卡洛斯三世大学信号理论与通信系创建的,旨在收集真实世界环境中的音频数据,特别是女性在日常生活中的情感和情境反应。数据集包含14名用户的音频、生理信号、GPS定位数据以及自我标注的情感和情境标签,总时长约为一周。数据集的创建过程涉及多模态数据的采集和标注,旨在通过自监督学习方法分析音频场景中的时空特征,特别是用于检测异常情境(如暴力环境)。该数据集的应用领域主要集中在情感计算和性别暴力风险预测。
The WE-LIVE dataset was developed by the Department of Signal Theory and Communications at Carlos III University of Madrid. It is designed to collect audio data from real-world environments, specifically capturing emotional and situational responses of women during daily life. The dataset includes audio recordings, physiological signals, GPS location data, and self-annotated emotional and situational labels from 14 users, with a total duration of approximately one week. The creation of this dataset involves the collection and annotation of multimodal data, with the objective of analyzing spatiotemporal features in audio scenes through self-supervised learning approaches, particularly for detecting anomalous situations such as violent environments. The primary application domains of this dataset are affective computing and gender-based violence risk prediction.




