lmms-lab/gigaspeech

该数据集是一个音频数据集，包含了多个类别的音频片段，如人物和博客、商业、非营利组织和激进主义等。每个音频片段都有详细的元数据，包括片段ID、演讲者、音频文件、开始和结束时间、音频ID、标题、URL、媒体来源、类别、原始完整路径和提示信息。数据集支持多个配置，并且提供了验证集和测试集。

This dataset is an audio dataset containing audio clips of various categories such as People and Blogs, Business, Nonprofits and Activism, etc. Each audio clip has detailed metadata including segment ID, speaker, audio file, start and end time, audio ID, title, URL, media source, category, original full path, and prompt information. The dataset supports multiple configurations and provides validation and test sets.