Hugging Face's logo

Datasets:
msingiai
/
eac-nsw-dataset

Modalities:
Audio
Text
Formats:
csv
Languages:
Swahili
Ganda
Kinyarwanda
Size:
1K - 10K
Tags:
text-to-speech
code-switching
non-standard-words
east-africa
swahili
luganda
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1

You need to agree to share your Hugging Face information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

EAC NSW Dataset

East African code-switching and non-standard-word text with matching synthetic speech.

Contents

  • data/eac-nsw-dataset.csv — 6,275 text rows.
  • audio/ — 6,275 WAV clips at 48 kHz; metadata in audio/manifest.json.

Audio: 11.1 hours total.

Languages: sw 1,941; en_ea 1,825; lg 961; rw 933; sheng 615.

Splits

  • data/train.csv — 5,020 rows.
  • data/test.csv — 1,255 rows.

80/20 random stratified split by language, category and domain (seed 1729); all languages, categories and domains are in test.

Access

This dataset is gated. Request access and wait for manual approval.

Notes

Synthetic speech; the text is not native-reviewed.
Downloads last month
109
Total file size:
3.84 GB