30 Jul 2025

African Next Voices Featured in Nature: Reshaping Speech Technology for Africa

Nature profiles the African Next Voices project, the largest multilingual speech data initiative in Africa β€” and DSFSI is proud to lead the South African leg. Learn how we’re making African languages AI-ready and inviting partners to build with us.

πŸ“° We’re in Nature!
πŸ“… Published: 29 July 2025
πŸ”— Full article on Nature.com

The African Next Voices (ANV) project β€” a pan-African initiative to address the deep underrepresentation of African languages in AI β€” has been profiled in Nature News. With over 9,000 hours of recorded speech in 18 languages across South Africa, Kenya, and Nigeria, this work is the most extensive open dataset creation effort of its kind on the continent.

β€œAI models are neglecting African languages β€” scientists want to change that.” β€” Nature, July 2025

DSFSI Leads the South African Leg πŸ‡ΏπŸ‡¦

At the Data Science for Social Impact group (DSFSI) at the University of Pretoria, we’re proud to be coordinating the South African arm of ANV, which focuses on seven languages:
isiZulu, isiXhosa, Sesotho, Setswana, Sepedi, isiNdebele, and Tshivenda.

Our work spans:

  • πŸŽ™οΈ Community-driven data collection and transcription
  • πŸ§ͺ Aligning data to key social sectors like education, health, and agriculture
  • πŸ“š Supporting downstream use in speech recognition, translation, and language models

Ethically Licensed, Openly Available πŸ“‚

All datasets are being released under open and permissive licensing that ensures public access β€” while requiring appropriate attribution to support transparency, research recognition, and responsible use. We believe that equitable AI starts with equitable data.

πŸ”— Explore the project and South African resources here:
πŸ‘‰ https://www.dsfsi.co.za/za-african-next-voices/

πŸ‘₯ Call to Action: Let’s Collaborate!

Are you a researcher, developer, civil society organization, or product team looking to work with speech data in African languages?

We want to hear from you!

πŸ“© Reach out via dsfsi.info@up.ac.za β€” let’s explore how you can use, contribute to, or extend this growing multilingual resource for Africa’s AI future.