A voice for 12 million speakers
“With speech technology, no one will be left behind.”
— Audace Niyonkuru, Digital Umuganda CEO
The problem Audace Niyonkuru faced
None of the big voice assistants understand a single native African language. For Kinyarwanda's 12 million speakers — many with low literacy or visual impairments — the entire voice-technology revolution simply didn't exist in their language.
What the project does about it
Mozilla's Common Voice crowdsourced more than 1,700 hours of donated Kinyarwanda speech into a public-domain dataset. Rwandan AI startup Digital Umuganda built speech-to-text and text-to-speech on it — including Mbaza, a chatbot that gives citizens government and health information in Kinyarwanda, spoken aloud.
Who this helped
Audace Niyonkuru, Digital Umuganda CEO — in their own words, above.