By:
Robert Kwaku Annor:
The Maseno Centre for
Applied Artificial Intelligence (MCAAI) has conducted a pre-visit to Suba ahead
of a main data collection exercise aimed at gathering quality speech data to
improve the representation of the Dholuo language in artificial intelligence
systems.
The visit was conducted by
the African Centre for Technology Studies (ACTS)-AI4D team leads and system
lead to better understand the community, engage Dholuo-speaking participants
and prepare them for accurate speech data collection.
During the exercise, Stanley
Oduor, a Masters student at Maseno University and Software Engineer at MCAAI,
introduced participants to the overall system and explained the importance of
quality training data. He demonstrated the difference between ideal data and
what is commonly described as “dirty data.”
The team also introduced
participants to the SmartVoice application, assessed their familiarity with the
system, gathered user acceptance feedback and identified potential challenges
ahead of the main data collection exercise.
According to MCAAI, the Suba
community was welcoming and willing to contribute to research aimed at
advancing the representation of Dholuo in AI.
Valary Otieno, a Masters
student at Maseno University and Machine Learning Engineer at MCAAI, together
with Jack Oraro, Software Engineer at MCAAI, supported the onboarding of users
to the SmartVoice system that will be used for data collection.
MCAAI said the exercise
forms part of its broader contribution to research focused on developing AI
systems that better represent African languages and communities.
The centre also acknowledged
Dr Vivian Oloo, Principal Investigator, Dr Lilian D. A. Wanzare,
Co-Investigator, and the entire MCAAI fraternity for their contributions to the
research.
