The Yoruba Corpus Initiative
Building the largest open-access linguistic dataset for Yoruba — tonal annotations, proverbs, oral literature, and spoken audio from native speakers across Nigeria and the diaspora. The dataset is designed to the quality standards required for training large language models, not merely for archival purposes.
Benin Bronze Digital Archive
High-resolution 3D scanning and provenance documentation of Benin Bronze works held in public and private collections globally — creating an authoritative, community-verified digital record of one of Africa's most significant artistic traditions. All documentation is conducted in partnership with Benin Kingdom cultural authorities.
Voices of the Sahel
An oral history recording programme across five Sahelian communities, capturing elder knowledge in Hausa, Fulfulde, and Kanuri — three languages critical to the region's cultural continuity and largely absent from global AI datasets. Recording is conducted by trained community members using field-grade audio equipment.
AfriSpeech Model
A speech recognition model trained specifically on African English and pidgin variants — addressing the significant accuracy gap that existing models show for West and East African accents in professional and institutional settings. Current benchmarks show a 40% improvement in word error rate over leading baseline models.
Great Zimbabwe VR Reconstruction
A photogrammetric and historical reconstruction of Great Zimbabwe at its architectural peak — building an immersive VR experience grounded in archaeological evidence and community consultation with Shona cultural custodians. The project is being developed in active partnership with the University of Zimbabwe.
AfriCorpus Open API
A public developer API providing access to our language models, cultural datasets, and translation tools — enabling researchers, educators, and builders across the continent to integrate African intelligence into their own products and research. Beta access planned for Q3 2025.