Kencorpus
lab · KenyaScores
1 product on the map — 1 open.
Openness
5 high confidence- license
- cc-by-4.0 (card metadata on all six repositories)
- access
- public (Hugging Face repositories ungated)
- dataset_card
- present (each repository has statistics, fields and usage)
All six repositories download without a gate under an attribution-only license. The paper calls it one of few public-domain corpora for these languages, though the cards use CC BY 4.0 rather than a public-domain dedication.
- https://arxiv.org/abs/2208.12081 recorded 2026-09-24
"Kencorpus is one of few public domain corpora for these three low resource languages".
- https://huggingface.co/api/datasets/Kencorpus/KenCorpus_text recorded 2026-09-24
"gated": false; cardData license "cc-by-4.0".
- https://huggingface.co/datasets/Kencorpus/KenCorpus_text/raw/main/README.md recorded 2026-09-24
Front matter "license: cc-by-4.0"; files per language and sources described.
Adoption
1 high confidenceHugging Face downloads summed over the text, audio and four task repositories that carry the Kencorpus paper link. Downloads count file fetches, not users.
- https://huggingface.co/api/datasets/Kencorpus/KenCorpus_audio recorded 2026-09-24
100 downloads in the trailing 30 days for Kencorpus/KenCorpus_audio
- https://huggingface.co/api/datasets/Kencorpus/KenCorpus_text recorded 2026-09-24
164 downloads in the trailing 30 days for Kencorpus/KenCorpus_text
- https://huggingface.co/api/datasets/Kencorpus/KenPOS recorded 2026-09-24
58 downloads in the trailing 30 days for Kencorpus/KenPOS
- https://huggingface.co/api/datasets/Kencorpus/KenSpeech recorded 2026-09-24
316 downloads in the trailing 30 days for Kencorpus/KenSpeech
- https://huggingface.co/api/datasets/Kencorpus/KenSwQuAD recorded 2026-09-24
132 downloads in the trailing 30 days for Kencorpus/KenSwQuAD
- https://huggingface.co/api/datasets/Kencorpus/KenTrans recorded 2026-09-24
112 downloads in the trailing 30 days for Kencorpus/KenTrans
Capability
1 medium confidenceKencorpus gives Swahili, Dholuo and Luhya a community-sourced text and speech corpus with tagging, question answering and translation sets, documented in a paper, with one community translation model trained on it. Most of its audio is untranscribed, and both its speech and its few million words of text are small beside English corpora of hundreds of thousands of transcribed hours and trillions of tokens, while WURA's African web corpus trained the AfriTeVa models.
- https://arxiv.org/abs/2208.12081 recorded 2026-11-12
Abstract: "5,594 items - 4,442 texts (5.6M words) and 1,152 speech files (177hrs)"; POS sets, 7,537 QA pairs and 13,400 translation sentences.
- https://arxiv.org/pdf/2208.12081 recorded 2026-11-12
Table 4.1 total row: 4,442 texts, 5,601,915 words, 1,152 speech files, 176:29:46; the STT proof of concept used "a collection of the speech corpus that is 27hrs 31min 50 sec".
- https://datasets-server.huggingface.co/size?dataset=Kencorpus/KenCorpus_audio recorded 2026-11-12
num_rows 915 in the default config.
- https://datasets-server.huggingface.co/size?dataset=Kencorpus/KenCorpus_text recorded 2026-11-12
num_rows 3787 in the default config.
- https://huggingface.co/api/models?filter=dataset:Kencorpus/KenCorpus_text recorded 2026-11-12
One model, Omballa/nllb-200-swa-dholuo, tagged with dataset:Kencorpus/KenCorpus_text.
- https://huggingface.co/datasets/Kencorpus/KenCorpus_audio/raw/main/README.md recorded 2026-11-12
"| **Total** | **1,152** | **176h 29m 46s** |"; the data fields are audio, language, genre, source, title, story_id and duration_seconds, with no transcript field.
- https://huggingface.co/datasets/Kencorpus/KenSpeech/raw/main/README.md recorded 2026-11-12
"| Total Duration | 27 hours 31 minutes 50 seconds |"; "| transcript | string | Transcription text |".
- https://huggingface.co/datasets/Kencorpus/KenSwQuAD/raw/main/README.md recorded 2026-11-12
"The dataset contains **7,506 question-answer pairs** derived from **1,441 unique Swahili contexts**".