Ellipsis Health 的美国英语对话语音语料库
收藏资源简介:
Ellipsis Health 的美国英语对话语音语料库是一个专有的数据集,包含10,932个独特的说话者,主要用于研究基于语音的抑郁症预测。数据集中的语音样本来自人机交互,每个会话平均包含354秒的音频,并附有PHQ-8抑郁症量表的自我报告结果。数据集的创建旨在通过大规模数据验证迁移学习在抑郁症预测中的效果,特别是在二分类和回归任务中的应用。
Ellipsis Health's American English conversational speech corpus is a proprietary dataset encompassing 10,932 unique speakers, primarily designed for research on speech-based depression prediction. The speech samples within the dataset originate from human-machine interactions, with each session averaging 354 seconds of audio, and are paired with self-reported outcomes from the PHQ-8 depression scale. This dataset was created to validate the effectiveness of transfer learning in depression prediction, specifically for binary classification and regression tasks, leveraging large-scale data.




