Atom Grants
Discover
  • Dashboard
  • Search
  • Deep Research
  • Favorites
  • Projects

    BENIN Fongbe Tonal ASR Challenge

    BENIN Fongbe Tonal ASR Challenge supports AI builders developing low-cost speech recognition that preserves Fongbe tones to expand access to voice-based public services, information and education.

    OverviewSimilar Grants
    OverviewSimilar Grants

    Funder: Zindi

    Due Dates: November 20, 2026 (Launch) | December 13, 2026 (Submission close) | December 13, 2026 (Submission selection close) | December 13, 2026 (Leaderboard reveal)

    Funding Amounts: $5,000 USD prize pool: $2,500 first place, $1,500 second place, $1,000 third place

    Summary: Build low-cost speech recognition models that transcribe spoken Fongbe into correctly spelled, fully tone-marked text.

    Key Information: Winning solutions must meet strict single-thread CPU inference limits and assign worldwide copyright in winning code to Zindi.


    Description

    This Zindi prize competition challenges AI builders to develop automatic speech recognition models for Fongbe, a tonal language spoken by around 2.3 million native speakers in southern Benin. Models must preserve tone marks, since a misplaced tone can change meaning. The goal is reliable speech technology for low-cost devices and local servers, expanding access to voice-based public services, information and education.

    The challenge is organised by Benin’s Ministry of Digital Transformation and Innovation with the Agency for Information Systems and Digital Affairs and the Institute for Inclusive Digital Africa. It forms part of the Gates Foundation-funded J’aime Ma Langue project. Participants work with mixed-quality speech recordings, including linguist-validated transcriptions and less reliable tone annotations. Winning approaches could inform speech recognition for other tonal languages in Benin and West Africa.


    Atom

    See the full grant listing

    Sign in to view full eligibility details, sources, similar grants, and AI-powered analysis.