A PhD student at the Kwame Nkrumah University of Science and Technology (KNUST), Ama Branoa Banful, is seeking to develop artificial intelligence (AI) speech recognition models capable of understanding people who switch between English and Ghanaian indigenous languages in everyday conversations.
The research aims to address the limitations of existing speech recognition systems, which are largely trained to process one language at a time and often struggle to understand speakers who mix two or more languages in a single sentence.
Ama Branoa Banful disclosed this during a courtesy call on the Ashanti Regional Minister, Dr Frank Amoakohene, at the Ashanti Regional Coordinating Council.
She explained that her interest in code-switching developed during her master’s research and would be expanded through her doctoral studies, with a focus on building a speech-to-speech system that could make digital technology more accessible to people who prefer speaking to reading or writing.
“For my master’s, I started looking into how I can incorporate code-switching, which is when you mix two or more languages in a given sentence,” she said.
She noted that code-switching is common in Ghana, where people frequently combine English with languages such as Fante and Pidgin in their daily conversations.
According to her, the next stage of her research would involve developing a speech-to-speech pipeline that can recognise and process such multilingual conversations more effectively.
She explained that the technology could help bridge the digital divide, particularly among people in rural communities who may find speaking more convenient than reading or writing.
“If you want someone in the village to benefit from what we are doing, we have to start getting into speech systems instead of just speech-to-text systems,” she said.
Banful indicated that existing AI models often struggle to follow conversations in which speakers alternate between languages because the systems are generally trained on individual languages rather than mixed-language speech.
She said her research would explore how different languages could be integrated into a single model without compromising the system’s ability to understand what a speaker is saying.
The research is expected to build on ongoing efforts by Ghanaian researchers and organisations to develop datasets in indigenous languages for use in AI applications.
During the engagement, the Minister for Communication, Digital Technology and Innovation, Samuel Nartey George, highlighted the availability of local-language datasets that could support Ama Banful’s work.
He referred to research collaborations involving Google and the University of Ghana, which supported the development of datasets covering seven indigenous Ghanaian languages, comprising approximately 5,000 hours of data.
He noted that the datasets were intended for public use and could be shared with researchers beyond the University of Ghana to advance language technology development.
The minister also referred to a broader collection of approximately 35,000 hours of Ghanaian indigenous-language data, describing it as valuable material for training and improving AI models.
He encouraged the researcher to take advantage of the available resources to strengthen her work and contribute to the development of speech technologies suited to Ghanaian and African languages.
The discussions also highlighted the importance of expanding language technology beyond Ghana to other African countries, where multilingual communication is common.
Researchers are increasingly exploring datasets in languages such as Hausa and Swahili to develop systems that can serve users across different linguistic communities.
Ama Banful’s proposed research could contribute to efforts to make AI-powered services more inclusive by enabling people to interact with digital platforms in the languages and communication styles they use in everyday life.
The initiative comes amid growing interest in developing locally relevant AI solutions that address language barriers and improve access to digital services across Ghana and the wider African continent.
