Google Research and the data-science platform Zindi ran a challenge to improve speech recognition for Lingala and Shona languages. The challenge drew 1,462 entrants from 100 countries, with 42 from Africa, who submitted 8,837 solutions.
The top three winners were Team Pasketti, Alban Nyantudre, and Abdourahamane Ide Salifou, with the top 3 scores being close. Nyantudre, a machine-learning engineer from Burkina Faso, stated that many people he grew up with speak their language perfectly but cannot read or write it.
The challenge used the WAXAL dataset, an open collection of African-language speech released by Google on 2 February 2026. The dataset covers languages spoken by more than 100 million people and is published on Hugging Face under Creative Commons licenses.
WAXAL is part of a wider effort to provide African languages with training data, including the release of CommonLingua, which covers 61 African languages. The Lingala and Shona results are a first test, and the next step is to see if the winning methods and data can be applied to other languages and services.
The WAXAL dataset was gathered with the help of several universities and organizations, including Makerere University, the University of Ghana, and the African Institute for Mathematical Sciences in Senegal. The data is published on Hugging Face under Creative Commons licences, which allow commercial reuse with attribution.
Millions of people speak Lingala and Shona fluently without ever having learned to read or write them. That makes voice the obvious way for them to use technology, and the hardest one to build, because most speech-recognition systems cannot understand either language.
Leave a Reply