• 120 hannotated audio (target)
  • 2languages in progress
  • 3pilot use cases

Why this project

Much of the population communicates orally first, in national languages. Digital services must be able to understand them.

Objectives

  • Build an open, annotated audio corpus
  • Train frugal speech-recognition models
  • Test use cases: farming information, health, public services

Features

  • Mooré and Dioula transcription
  • Voice Q&A over a document base
  • Runs on a modest server

A data project? Let's talk.

Free first conversation, reply within one business day.

Contact us

Newsletter