AI For Africa, By Africans
African Languages Lab is Africa's leading language AI research company. We build the datasets, models and translation systems that African languages have been missing, and deploy them for enterprises, governments and innovators working across the continent.
Most African languages have no presence in the systems that increasingly decide how people learn, work and are served. African Languages Lab began as a research effort to understand that gap, and now builds the datasets, models and infrastructure beneath African language AI.
To build the world's most powerful African language AI, grounded in research and deployed at enterprise scale, so that African languages work in technology and African voices shape what gets built next.
What we do
We collect the data, do the research, build the models, and put them into production.
We collect the data
Speech and text gathered and validated across the continent through All Voices and our expert linguist networks. Over 100 billion curated tokens and 13,000+ hours of validated speech.
We do the research
Low-resource training, speech recognition, speech synthesis and evaluation. Peer-reviewed and presented at venues including ACL.
We build the models
Multilingual, multimodal systems covering 30+ African languages, from machine translation to speech to reasoning.
We deploy at scale
Low-resource training, speech recognition, speech synthesis and evaluation. Peer-reviewed and presented at venues including ACL.
Our Footprint
3
3
2
1
2
3
2
1
+
+
+
+
+
01
3
3
2
1
2
3
2
1
02
02
Curated text tokens
Text collected and cleaned across African languages, the largest such corpus in existence.
Hours of validated speech
Recorded and checked by native and fluent speakers across the continent.
Languages collected
From widely spoken languages to those with no prior digital record.
2
3
2
1
+
+
+
+
+
01
3
3
2
1
%
%
%
%
02
%
%
%
%
02
Languages in production
Live in Mansa today, across translation, transcription, speech, interpretation and chat.
Languages added to Common Crawl
Contributed to one of the industry's largest open training sources.
Research institution collaborators
Collaborators including UCLA, Georgia Tech, Cornell, Columbia and the University of Cape Coast.