zalozbadev/hsb_audio_corpus

这是一个包含上索布语语音录音和转录的数据集。多位演讲者为此数据集贡献了他们的声音。音频文件存储在sig文件夹的子文件夹中，相应的书面文本可以在trl文件夹中相同路径找到。子文件夹的结构如下：sig/资源ID/演讲者ID/录音会话/files.wav，相应的文本文件为trl/资源ID/演讲者ID/录音会话/files.trl。不同资源中匹配的演讲者ID表示是同一位演讲者。此外，从转录中生成了必要的metadata.csv文件，以便在数据集查看器中显示转录内容，并方便使用数据集。

This is a collection of speech recordings in Upper Sorbian, contributed by several speakers. Audio files are stored in subfolders of the sig folder, and the corresponding written text can be found in the trl folder at the same path. The structure of the subfolders is as follows: sig/ID_of_resource/ID_of_speaker/recording_session/files.wav, and the corresponding text file is trl/ID_of_resource/ID_of_speaker/recording_session/files.trl. Matching speaker IDs within different resources indicate the same speaker. Additionally, a necessary metadata.csv file has been generated from the transcriptions to display the transcripts in the dataset viewer and to facilitate easy usage of the dataset.