README.md

metadata

license: other
datasets:
  - reazon-research/reazonspeech
task_categories:
  - automatic-speech-recognition
language:
  - ja
pretty_name: ReazonSpeech
size_categories:
  - 10M<n<100M

Reazon Speech v2 dataset mirror

Original Dataset Source

• Hugging Face Dataset Page: reazon-research/reazonspeech
• Project Page: Reazon Research

License

This dataset is a mirror of the original Reazon Speech v2 dataset, but on 🤗 server (so may be faster). This dataset is licensed under the CDLA-Sharing-1.0. The original dataset comes with the following restriction:

TO USE THIS DATASET, YOU MUST AGREE THAT YOU WILL USE THE DATASET SOLELY FOR THE PURPOSE OF JAPANESE COPYRIGHT ACT ARTICLE 30-4.

This mirrored dataset is redistributed under the same CDLA-Sharing-1.0 license and the restriction.

Disclaimer

TO USE THIS DATASET, YOU MUST AGREE THAT YOU WILL USE THE DATASET SOLELY FOR THE PURPOSE OF JAPANESE COPYRIGHT ACT ARTICLE 30-4.

Notes

• No modifications have been made to the original dataset in this mirror.
• All credits for the dataset creation go to the original creators at Reazon Research.

Dataset Format

Audio files are available in FLAC format, sampled at 16000 hz.

• all.tsv: TSV file containing the audio file paths and their transcriptions, with the following format:

000/0123456abcdef.flac    書き起こしをお伝えします。
01f/987654321fdeb.flac    これは音声データです。
...

• metadata.csv: CSV file containing the same info as all.tsv:

file_name,transcription
data/000/0123456abcdef.flac,書き起こしをお伝えします。
data/01f/987654321fedc.flac,これは音声データです。
...

• data/: Directory containing the audio files. There are 4096 tar files data/001.tar -- data/fff.tar containing the audio files.

Structure of the repository:

.
├── all.tsv
├── metadata.csv
└── data
    ├── 001.tar
    ├── 002.tar
    ├── 003.tar
    ├── ...
    └── fff.tar