Services & Offerings
MbazaNLP provides open-source NLP infrastructure and community resources for Kinyarwanda and East African languages. All tools are freely available under open licences.
Machine Translation
Open models for English–Kinyarwanda translation across general, education, and tourism domains. Fine-tuned on NLLB-200.
Browse models →Speech Recognition
ASR models for Kinyarwanda using Whisper, NeMo Conformer, and Coqui STT. Multilingual model covers Kinyarwanda, Swahili, and Luganda.
Browse models →Text-to-Speech
TTS synthesis model producing natural Kinyarwanda speech, suitable for accessibility and voice interface applications.
Browse models →Open Datasets
Parallel corpora and benchmark datasets for Kinyarwanda NLP research, openly licensed for academic and commercial use.
Browse datasets →Research Collaboration
MbazaNLP welcomes research partnerships with universities, institutes, and NGOs working on African language technology. We can collaborate on:
- Joint model development and evaluation
- Data collection and annotation sprints
- Grant applications (Lacuna Fund, AI4D, Mozilla)
- Co-authoring research publications
Training & Workshops
The community organises technical workshops and sprints for developers and researchers in Rwanda and the wider region. Topics include:
- Fine-tuning language models for low-resource languages
- Data collection and quality assurance for NLP
- Deploying AI models in production (APIs, demos)
- Contributing to open-source AI projects