I Built a Voice API So My Mother Could Book a Taxi in Lomé(Togo) (kuma.compeel.com)

KUMA has developed a voice API designed to bridge the language gap for Mina and Ewe speakers in Lomé, Togo, allowing them to use ride-hailing and delivery apps in their native tongues. The newly launched API processes audio input to discern user intentions—such as booking a ride or placing an order—facilitating seamless interactions with technology that previously lacked support for these local languages. This innovation addresses a significant barrier for millions who are familiar with these languages but have been underserved by existing applications. The KUMA voice API operates through a straightforward three-step process: users authenticate with credentials, obtain a token, and then submit audio for analysis. The system returns a structured intention output, complete with high confidence levels and relevant entities, which can be used directly by applications. For instance, a spoken command such as "Moudji be mayi asigamé" translates to a 94% confident intention to order a ride to Grand Marché. This development not only empowers local users but also underscores the growing importance of inclusivity in AI technologies, highlighting the necessity of language diversity in user experience and the potential for broader application in other underrepresented languages.