Browse Datasets

Kencorpus

Kenyan Languages Corpus: Swahili, Dholuo, Luhya

machine-translation pos-tagging question-answering 5.6M words / 177 hours CC BY 4.0
Access Dataset Read Paper

Description

A text and speech corpus for three Kenyan languages (Swahili, Dholuo, Luhya): 5.6M words and 177 hours of speech, plus 7,537 Swahili QA pairs, 13,400 translation sentences, and POS-tagged sets. The Maseno project homepage is offline; data is mirrored on Harvard Dataverse.

Quick Start

Python 
# Download from Harvard Dataverse (CC BY 4.0):
#   https://doi.org/10.7910/DVN/6N5V1K
# text, QA, translation and POS files plus speech wavs

Project Ideas

Dholuo-Swahili MT

Build a Dholuo-to-Swahili translation transformer.

Intermediate

Swahili STT

Develop a Swahili speech-to-text pipeline and measure word error rate.

Intermediate

Ethical Considerations

  • Indigenous languages carry cultural heritage; develop with native-speaker consultation.