Professor Cathy Holloway from GDI Hub highlighted the importance of linguistic diversity in AI development, stating that this collaboration will not only enhance accessibility in Ghana but also provide insights for global AI advancements.

This project builds on the existing Project Relate app, which supports communication for individuals with non-standard speech globally, and aims to expand its capabilities to include more languages and regions.

Specifically, the initiative will develop localized ASR models to improve AI-driven speech recognition technology for five major Ghanaian languages: Akan, Ewe, Dagaari, Dagbani, and Ikposo.

The University of Ghana has teamed up with Google Research Africa and UCL's Global Disability Innovation Hub on an innovative AI project called 'tɛkyerɛma pa,' aimed at enhancing communication for individuals with non-standard speech in Ghana.

With a grant of $40,000 from Google as part of the AT2030 programme, this initiative will focus on developing automatic speech recognition (ASR) models tailored for local languages.

Katrin Tomanek, a Senior Research Scientist at Google Research, has pledged to provide technical support and share insights from Project Euphonia to assist local researchers.

The focus will be on accurately interpreting diverse speech patterns, promoting inclusivity in communication for individuals with conditions such as cerebral palsy, ALS, and stroke.

The initiative embodies the theme 'AI by Africa for Africa,' showcasing the potential for locally led innovation to have a significant global impact.

The project was officially announced during a tech talk and press event held at the Google Ghana office.

Adwoa Amponsah Dapaah, a lawyer with cerebral palsy, emphasized that this initiative will foster inclusion for all individuals with speech impairments, regardless of their educational background.

As part of the project, local speech therapists will collect diverse speech samples in languages like Akan, Ewe, Ikposo, Dagbani, and Dagaare, addressing real-world accessibility barriers.

Professor Isaac Wiafe, the project lead, noted that initial data collection will focus on non-standard speech in the Akan language, particularly the Asante dialect, starting from Kumasi.