AMI corpus的咳嗽事件重新标注

AMI corpus的咳嗽事件重新标注数据集由都柏林理工大学生物医学研究组创建，包含1369个自然发生的咳嗽声音样本。该数据集通过对原始AMI corpus中的咳嗽事件进行重新标注而生成，解决了原有标注中存在的开始和结束时间不准确及多个连续咳嗽被标注为单一事件的问题。数据集的创建过程涉及使用专用GUI工具进行音频导入和新时间戳的记录。此数据集主要应用于机器学习领域，用于开发和测试与咳嗽声音相关的音频事件检测算法，旨在提高咳嗽声音分析和检测的准确性和可靠性。

The cough event re-annotated dataset of the AMI corpus was created by the Biomedical Research Group of Dublin Institute of Technology, containing 1369 naturally occurring cough sound samples. This dataset is generated via re-annotating the cough events in the original AMI corpus, which resolves two key issues in the original annotations: inaccurate start and end timestamps, and multiple consecutive coughs being labeled as a single event. The dataset creation process involved using a dedicated GUI tool for audio import and recording of new timestamps. This dataset is primarily utilized in the machine learning field for developing and testing audio event detection algorithms related to cough sounds, with the goal of improving the accuracy and reliability of cough sound analysis and detection.