JSUT (Japanese Speech Corpus) - Test Subset
A test subset of the JSUT corpus containing 500 Japanese utterances from the basic5000 dataset (BASIC5000_4501-5000).
Dataset Description
- Language: Japanese (ja)
- Size: 500 audio clips
- Format: WAV audio (48kHz) + text transcription
- Source: JSUT Corpus
- Subset: basic5000 (utterances 4501-5000)
- License: CC-BY-SA 4.0 (see LICENCE.txt)
Dataset Structure
jsut_ver1.1/
└── basic5000/
├── wav/ # WAV audio files (500 files, 48kHz)
├── transcript_utf8.txt # Transcriptions
└── recording_info.txt # Recording dates
File Formats
transcript_utf8.txt
BASIC5000_4501:だが、エーアイセンター稼動を快く思わない...
BASIC5000_4502:また、サイコロに姿を変えることもでき...
...
recording_info.txt
BASIC5000_4501.wav 12/19/2016
BASIC5000_4502.wav 12/19/2016
...
Usage
Loading with Hugging Face Datasets
from datasets import load_dataset
# Load from local path
dataset = load_dataset("audiofolder", data_dir="jsut_ver1.1/basic5000")
# Or after uploading to Hugging Face Hub
dataset = load_dataset("your-username/jsut-basic5000-test")
Loading with Python
import pandas as pd
# Read transcripts
transcripts = {}
with open("jsut_ver1.1/basic5000/transcript_utf8.txt", "r", encoding="utf-8") as f:
for line in f:
utt_id, text = line.strip().split(":", 1)
transcripts[utt_id] = text
print(transcripts["BASIC5000_4501"])
Citation
@article{sonobe2017jsut,
title={JSUT corpus: free large-scale Japanese speech corpus for end-to-end speech synthesis},
author={Sonobe, Ryosuke and Takamichi, Shinnosuke and Saruwatari, Hiroshi},
journal={arXiv preprint arXiv:1711.00354},
year={2017}
}
Acknowledgments
This dataset is derived from the JSUT corpus created by:
- Ryosuke Sonobe (University of Tokyo)
- Shinnosuke Takamichi (@forthshinji, University of Tokyo)
- Hiroshi Saruwatari (University of Tokyo)
The full corpus is available at: https://sites.google.com/site/shinnosuketakamichi/publication/jsut