Regional Center
Center for African Studies
CAS supports language learners with resources that connect language study to culture, history, and current events across the continent.
Current Projects
Tigrinya LLM, TTS & STT Initiative
A community-driven initiative to build a Tigrinya-specialized language model, text-to-speech voice, and speech-to-text system, trained on data collected and verified by native Tigrinya speakers. Tigrinya (~9-10M speakers, Eritrea and Ethiopia's Tigray region) has almost no dedicated learner tooling and only a thin layer of academic NLP resources that have never been connected to an actual product. This project fine-tunes existing open models (an open-weight LLM via QLoRA, Piper for TTS, Whisper for STT) on a growing, register-and-dialect-tagged sentence dataset built by a volunteer contributor group, rather than training from scratch.
Volunteer for this project