Low-resource
22m
speakers, medium confidence estimate
59
AI readiness, weighted toward speech
3
dialects tracked as separate collection targets
A large diaspora is the unusual feature: remittance and telehealth demand sits partly outside the continent.
Capability coverage
Some open data and big-tech ASR, but no production TTS. Demand exists; quality does not.
The facts
Dialects
Dialect is self-declared at contribution and then validated against gold-standard clips. Which dialects people turn up to record in is a better demand signal than any desk ranking.
Open corpora that cover it
Voices
No voice yet. Studio recording is the scarcest asset in this whole field — the largest open African corpus holds 11,000 hours and about twenty of them are studio-grade — so voices follow demand rather than leading it.
Same family
Related
A Yoruba model helps Igbo less than the shared border suggests, and a Hausa model transfers usefully across Chadic languages in three other countries. These share Afroasiatic with Somali.
The best cross-border expansion asset in West Africa — one language investment reaches Niger, Chad, Ghana and northern Cameroon. Also the one West African language ElevenLabs does cover, so the deficit here is smaller than it looks.
65 AI readiness
88m
A non-Latin script, so the text pipeline is a different problem from every other language on this list. Big speaker base, difficult FX and regulatory environment.
49 AI readiness
57m
The largest African language with effectively no commercial speech support anywhere. Its Latin orthography was adopted in 1991, so the text pipeline is cheap — which makes the absence of coverage harder to explain than for Amharic next door.
35 AI readiness
37m
Take an API key and make the call from your own code, or start in the playground and see what the model actually does with your text.
