Built for African Languages, Trained on African Voices
Built for African Languages, Trained on African Voices
Automatic Speech Recognition (ASR)
Khaya AI’s ASR tool converts speech into text in 32 African languages. It can be used to transcribe audio, process voice inputs, and support applications that require speech-to-text capabilities.
Supported languages include: Adangme, Akuapem Twi, Asante Twi, Dagbani, Dagaare, Ewe, African English, Fante, French, Ga, Gonja, Gurene, Hausa, Igbo, Kasem, Kikuyu, Konkomba (Likpakpaanl), Konkomba (Likoonli), Krio, Kusaal, Luo, Mampruli, Mende, Meru/Kimeru, Nzema, Pidgin, Shona, Swahili, Temne, Wali, Wolof, Yoruba
What Khaya Offers
Accurate Transcription
Khaya's ASR systems include models with Word Error Rates as low as 5%, and some achieving even 2%. The models continue to improve as we augment data and refine training pipelines.
Real‑World Integration
Use Khaya ASR to control smartphones or smart‑home devices via voice, generate meeting notes, or power voice search in your apps.
Expansion Roadmap
Swahili and Hausa ASR models are planned, along with additional Ghanaian languages . New languages are prioritised based on community need and data availability.

