Sam George Backs Indigenous-Language AI Research with 35,000 Hours of Ghanaian Data

The Ministry of Communication, Digital Technology and Innovations has highlighted the potential of approximately 35,000 hours of Ghanaian indigenous-language audio data to accelerate the development of speech recognition models and local artificial intelligence tools.

The Minister, Samuel Nartey George, pointed to research underway at the Kwame Nkrumah University of Science and Technology (KNUST) where doctoral researcher Ama Branoa Banful is building natural language models specifically capable of processing English-Twi code-switching.

"The availability of approximately 35,000 hours of Ghanaian indigenous-language data presents an opportunity for researchers to build AI models that respond more effectively to the needs of local users," the Minister stated during discussions on local AI innovation.

Most commercial speech-to-text systems encounter severe accuracy degradation when speakers blend English and Ghanaian languages within natural conversation. The research aims to make voice interfaces, automated transcription, and customer support technologies directly accessible to multi-lingual Ghanaian speakers.

The initiative forms part of broader efforts to ensure African linguistic representation in foundational AI models and empower local software developers.