yoad/crowd-transcribe-v5-xs

该数据集包含了音频文件及其对应的文本信息，音频文件可能经过重录，每个音频文件都由特定的转录者转录。数据集还包含了音频的其他属性，如时长、是否为外语、是否嘈杂等。此外，还记录了用户与音频互动的统计数据，如播放次数、暂停次数等。数据集分为训练集和测试集两部分。

The dataset consists of audio files and their corresponding text, the audio files may have been retranscribed, and each audio file is transcribed by a specific transcriber. The dataset also includes other attributes of the audio, such as duration, whether it is a foreign language, whether it is noisy, etc. In addition, it records the statistical data of user interaction with the audio, such as the number of plays, pauses, etc. The dataset is divided into two parts: training set and test set.