The University of Ghana, in partnership with Google Research Africa and the Global Disability Innovation Hub (GDI Hub) at University College London (UCL), has launched an AI-driven initiative called “tɛkyerɛma pa” (meaning “Good Tongue”). This project is designed to support better communication for individuals with unique speech patterns.
Aiming to enhance speech recognition technology for inclusivity, the initiative focuses on five prominent Ghanaian languages: Akan, Ewe, Ikposo, Dagbani, and Dagaare. This marks a significant advancement in AI-powered communication tools, ensuring that people with non-standard speech patterns are better understood and can engage more fully in their communities.
Around the world, individuals with non-standard speech—resulting from conditions such as cerebral palsy, ALS, stroke, Down syndrome, and Parkinson’s disease—often face major communication barriers.
Existing speech recognition technologies typically have difficulty accurately processing non-standard speech, which restricts their usability and impact.
The tɛkyerɛma pa project seeks to bridge this gap by training AI models to better recognize these unique speech patterns in Ghana’s local languages. This initiative aims to improve accessibility and create a more inclusive experience for Ghanaians with non-standard speech, enhancing the effectiveness of speech recognition for all.
At a press briefing and tech discussion held at Google Ghana’s office, Professor Isaac Wiafe, Project Lead and Associate Professor at the University of Ghana, highlighted the importance of Project tɛkyerɛma pa in fostering inclusive communication, saying, “No one should be left behind, and we are excited to lead this vital initiative.”
He emphasized that developing localized AI solutions could make effective communication accessible to more Ghanaians, regardless of speech patterns.
Funded by a $40,000 grant from Google, Project tɛkyerɛma pa benefits from UCL’s Global Disability Innovation (GDI) Hub’s expertise through the AT2030 program. Katrin Tomanek, Senior Research Scientist at Google Research, shared her enthusiasm for the collaboration, stating that Google is dedicated to supporting African AI research to ensure that advances benefit everyone, regardless of how they speak. The project includes the creation of the first open-source dataset for non-standard speech in Ghanaian languages.
Professor Cathy Holloway, Academic Director of the GDI Hub, underscored the importance of extending AI-powered speech recognition to additional languages. This expansion, she noted, is essential for enabling individuals with non-standard speech to interact more effectively with others and with devices, integrating better into daily life.
SOURCE: http://dew360.net
Join our WhatsApp channel: https://whatsapp.com/channel/0029VakDz4u9RZATWh53yC1a
