Mubarak Muhammad, a student and machine learning engineer, has emerged as the winner of the TechCabal Ewè audio classification challenge. The top three winners were announced at the recently concluded Moonshot by TechCabal conference held in Lagos themed “Building for the World.”.
The TechCabal Ewè audio classification challenge was organised in partnership with Zindi, the data science competition platform, and Umbaji as data partners. The competition was designed to address a social problem while furthering linguistic diversity in AI development. Competition participants developed machine learning models to classify audio recordings of basic commands spoken in Ewè. With over 5 million Ewè speakers, this challenge will contribute to voice assistance technology for visually impaired Ewè speakers.
Mubarak Muhammad (ML_Wizzard) clinched the $500 first prize for his high-efficiency mode based on the modified version of MatchboxNet. He incorporated several key architectural improvements that enhanced its performance for Ewè language command recognition. The model’s success in the TechCabal Ewè Audio Translation Challenge demonstrates its effectiveness for real-world applications in African language technology, particularly for assistive technologies for visually impaired Ewè speakers.
The second $300 prize went to Victor Olufemi (Professor) a data scientist.
The third prize of $200 went to Yves Ngoy (nymfree), the lead SI engineer at US-based Altair, a software development company.
