Dataset Card for ReazonSpeech

Dataset Summary

ReazonSpeech is a large audio corpus collected from Japanese TV programs.

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

Languages

Japanese

Dataset Structure

Data Instances

The following shows an example dataset record:

{
    'name': '000/0000000000000.flac',
    'audio': {
        'path': '/path/to/000/0000000000000.flac',
        'array': array([ 0.01000000,  ...], dtype=float32),
        'sampling_rate': 16000},
    'transcription': '今日のニュースをお伝えします。'}
}

Data Fields

Field	Type	Desc
name	string	An unique id for the audio file
audio	dict	A dictionary containing the file path, the decoded audio, and the sampling rate
transcription	string	A text transcription of the audio data

Data Splits

Split	Size	Desc
all	>1TB	Contains all the dataset
small	350MB	Small subset for example purposes (default)

Additional Information

Dataset Curators

Reazon Human Interaction Laboratory

Licensing Information

CDLA-Sharing-1.0

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