COUGHVID

COUGHVID 数据集提供了超过 20,000 个众包咳嗽记录，代表了广泛的受试者年龄、性别、地理位置和 COVID-19 状态。首先，使用开源咳嗽检测算法过滤数据集。其次，经验丰富的肺科医生标记了 2,000 多条记录以诊断咳嗽中存在的医学异常，从而贡献了现有最大的专家标记咳嗽数据集之一，可用于大量咳嗽音频分类任务。

The COUGHVID dataset provides over 20,000 crowdsourced cough recordings, covering a broad spectrum of subjects' ages, genders, geographic locations, and COVID-19 infection statuses. First, the dataset was filtered using an open-source cough detection algorithm. Second, more than 2,000 recordings were annotated by experienced pulmonologists to identify medical abnormalities in cough sounds, resulting in one of the largest existing expert-annotated cough datasets that supports a wide range of cough audio classification tasks.