Sudanese Arabic
/apd/Status: BuildingOur first dataset — everyday speech across dialects, recorded by voice.
Speakers: ~30M speakers
Held privately
Datasets
Our data
Speech recordings in Sudanese Arabic, Zaghawa and Fur, recorded and checked by native speakers. Here is where each one stands.
Where things stand
These are real figures, not targets. The recordings are held privately: they are the foundation's own data for training and measuring its models, and they are not an open dataset. Sharing any of it beyond the foundation would be a separate, deliberate decision, taken only with the consent of the people who recorded it.
Fur
/fvr/Status: EarlyA tonal language of Darfur; collection design with native speakers underway.
Speakers: ~1M speakers
Held privately
Zaghawa (Beria)
/zag/Status: EarlyBeria, spoken across Sudan and Chad; script and tone documentation in scope.
Speakers: ~350K speakers
Held privately