Dataset Viewer
Auto-converted to Parquet Duplicate
The dataset viewer is not available for this split.
Server error while post-processing the rows. Please report the issue.
Error code:   RowsPostProcessingError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Yoruba Speech Dataset (OpenSLR 86 - Female Speakers)

A high-quality Yoruba speech dataset from female speakers, sourced from OpenSLR 86.

Usage

from datasets import load_dataset

dataset = load_dataset("oademuwagun/yoruba-speech-female")

Dataset Structure

├── metadata.csv
└── wavs/
    ├── yof_06136_00616155826.wav
    ├── yof_04310_00085429033.wav
    └── ...

Metadata Format

Column Description
file_name Path to audio file
transcription Yoruba text with full diacritization

Yoruba Diacritics

The transcriptions include proper Yoruba diacritical marks:

  • Tone marks: acute (́) for high tone, grave (̀) for low tone
  • Underdots: ẹ, ọ, ṣ

Source

This dataset is derived from OpenSLR 86, a crowdsourced high-quality Yoruba speech dataset developed by Google.

Original Paper

Gutkin, A., Ha, L., Banski, P., et al. (2020). "Developing an Open-Source Corpus of Yoruba Speech." Proc. Interspeech 2020.

License

CC-BY-SA-4.0 (Creative Commons Attribution-ShareAlike 4.0 International)

Downloads last month
20