Centre for Digital Language Inclusion in Ghana

During February 2025 , CDLI’s Professor Cathy Holloway, Dr Richard Cave and Dr Giulia Barbareschi worked with the team from University of Ghana and Talking Tipps Africa to help collect datasets of impaired speech in two Ghanaian languages, Akan and Ga. These datasets will are being used by CDLI and University of Ghana to build automatic speech recognition (ASR) models that can help people living with impaired speech to be better understood using the languages they prefer to speak in. The ASR models will be available for the GDI-Hub funded Hackathon in Accra during May 2025, that is open to all universities in Ghana and beyond. The teams will ideate and develop functional solutions for everyday communication for people with impaired speech using this AI-driven technology.

“AI-driven inclusive communication technologies are essential in ensuring AI is accessible to all”

Richard Cave PhD 

CDLI is building these open-source tools on top of the learnings from Google’s work on non-standard speech and GDI Hub’s UK aid-funded AT2030 Programme, which first investigated how this technology could be used by Africans and beyond the English language. CDLI is ensuring this technology is available to those who would most benefit: non-English-speaking populations in African countries, who often have the least access to traditional assistive tech and speech and language therapy. 

Why does this matter? 

Millions of non-English-speaking people in Africa lack access to traditional assistive technology and speech therapy. This initiative will bridge that gap by:
1. Developing inclusive solutions for people with communication difficulties
2. Expanding data collection of non-standard speech across native African languages (with support from Google.org)
3. Creating open-source datasets & knowledge banks to empower local innovators to develop speech tech for their own communities.

This is a critical step toward inclusive AI—ensuring everyone benefits and no one is left behind.