Tech

Google, and UG collaborate to develop speech recognition for Ghanaian languages

For Ghanaian language preservation and digital inclusion, Google Research Africa, in partnership with the University of Ghana (UG), University College of London’s (UCL) Global Disability Innovation has launched a project to develop speech recognition technology for indigenous Ghanaian languages.

The initiative, named Tekyerema Pa, aims to enhance communications for individuals with non-standard speech patterns due to conditions such us Down Syndrome, Stroke, Cerebral Palsy, and Parkinson’s Diseases.

 

The project, Tekyerema Pa focuses on creating a robust language database for popular Ghanaian languages such as Asante (twi), Ewe, Dagaare and Dagbani. Speech recognition technology enables computers and devices to understand and process spoken language, paving the way for applications in translation for people in speech impairment, making it easy to communicate through digital tools accessible to Ghanaians in local languages.

 

Speaking to the media on Friday, Professor Isaac Wiafe, project leader and associate professor at University explained the progress of the project. In his address, he stated that Tekyerema Pa is a training AI model which will better the understanding of unique pattern in our local languages.

“By developing localized AI Solution, we can assure that more Ghanaians will have the opportunity to communicate effectively, regarded of their speech pattern”

 

The project’s development team is using a combination of artificial intelligence (AI) and machine learning algorithms to analyze phonetic patterns and regional dialects, ensuring that the technology recognizes and accurately interprets the unique characteristics of each language. They are also crowdsourcing voice data from native speakers to build a comprehensive database, helping the model understand variations in pronunciation and vocabulary.

Google representatives, Ms. Katrin Tomanek, a Senior Research Scientist at Google Search emphasized that the Tekyerema Pa project builds on continuous research aimed at making Automatic Speech Recognition (ASR) more accessible.

The project has already made strides in producing beta versions of the technology, which will soon be made available for public testing. Once finalized, this tool could have significant implications for Ghana healthcare, and government sectors, offering a digital solution for sectors that rely heavily on clear communication in local languages.

 

Tekyerema Pa is supported by a $40,000 grant from Google, with research expertise provide through AT2030 program, an assistive technology initiative from the UCL’s GDI Hub.

 

Amnewsagency | Reporters | Accra.

Amnewsworld

We gather, verify, distribute news and information to the general public, media outlets, and businesses. Our channels shares politics, arts & culture, sports, business, Lifestyle, science, technology, health, education, history and environmental news across the globe.

Recent Posts

132-Year-Old Shipwreck of the Western Reserve discovered in Lake Superior

For over a century, the depths of Lake Superior hid the wreckage of the Western…

34 minutes ago

Trump raises Steel and Aluminum tariffs to 25%, betting on U.S. manufacturing boom

In a bold move that could reshape global trade and rattle markets, President Donald Trump…

41 minutes ago

Uganda deploys troops to South Sudan amid rising tensions between Kiir and Machar

In a move signaling deepening instability in South Sudan, Uganda has deployed an undisclosed number…

16 hours ago

The silencing of African Voices: Who owns the right to tell African Stories?

For decades, the global media landscape has operated under an unspoken rule: Western narratives about…

1 day ago

Elon Musk claims X outage was due to a massive Cyberattack

Hours after widespread outages left thousands of users unable to access X, the social media…

1 day ago

U.S. Stock Market Sell-Off Deepens amid Tariff Concerns

Wall Street Takes a Hit as Uncertainty Over Trump’s Policies Grows The U.S. stock market…

1 day ago

This website uses cookies.