Advancing Support for African Languages

"The future is already here - it's just not very evenly distributed"

We're researching Natural Language Processing (NLP) models to bring underrepresented African languages to the technological cutting edge, making information accessible to everyone.

Explore Our Models

Our Models

Our open-weight, state-of-the-art automatic speech recognition models trained on African languages, available for free download and use.

whisper-large-v3-shona-asr

sna ASR

Automatic Speech Recognition model for Shona, based on Whisper v3 and fine-tuned on 12 hours of audio. Can handle transcription of Shona audio to text.

Model Parameters 2B
WER 23.2%

whisper-large-v3-lingala-asr

lin ASR

Automatic Speech Recognition model for Lingala, based on Whisper v3 and fine-tuned on 12 hours of audio. Can handle transcription of Lingala audio to text.

Model Parameters 2B
WER 14.8%