Google Launches New Speech Dataset to Advance African Languages in AI Technology.
· AI Study
Google and African researchers launch open-access WAXAL dataset to expand AI voice technology in 21 African languages.
Google, in collaboration with African research institutions, has announced the release of an open-access speech dataset called WAXAL . This dataset includes 21 Sub-Saharan African languages, such as Swahili, Hausa, Yoruba, and Luganda. The primary goal of this initiative is to bridge the digital divide and benefit over 100 million speakers who were previously excluded from voice technologies due to a lack of high-quality language resources.
Funded by Google and developed over three years, the project features 1,250 hours of transcribed natural speech and high-quality studio recordings designed to create superior synthetic voices. Local institutions, including Makerere University and the University of Ghana, led the data collection efforts. Crucially, ownership of the data remains with these African institutions, allowing researchers and students to develop technological solutions in their own languages and local contexts without depending on foreign corporations.
As reported by AI Africa, the WAXAL dataset is now available to any researcher or entrepreneur. It is already sparking innovation in key sectors such as healthcare, education, and agriculture, opening significant doors for Africans to engage with technology in their native tongues.
Source: Ethiopian Artificial Intelligence Institute