Who is Kevin?
Kevin is a researcher in technology, Lecturer in computer science and also a Manager and contributor to open source projects such as Common Voice which is a Mozilla project.
Tell us how you got involved with this Common Voice project?
I found out about the Common Voice project in 2019 through my PhD degree administrator Dr. Tom Denton from Google. We discussed a lot about how machines can be trained to work as human beings and that is how Common Voice emerged. The focus of the discussion was to develop a sound recognition engine and make it an open source.
Looking back since when you started contributing to the Common Voice Kiwahili project till now, what can you say have been your successes?
Great success for me has been to be able to revive the Kiswahili community especially here in Nairobi. This is after the Kiswahili community's long absence, but after meeting people like George Githuma, Eunice Manyasi, Hildagard Msuya, Victoria Nyamai, Joan Wambui and others who have volunteered to contribute to this volunteer work and want to see the work grow bigger and have a positive impact in the future.
What challenges have you faced as you look back on this journey of Common Voice?
I will call them positive challenges, and these include spending more time away from my daily activities to sustain contributions to the Common Voice platform. Communities sometimes shrink as contributors become more involved with other things. Persuading people to join a project that has no financial reward has also been weighing on my mind.
Speaking of challenges, how did you manage to overcome these challenges?
One of the ways I have used to solve these challenges, is to find and work with contributors who have a shared vision with this project and also those who are willing to do volunteer work. Through these contributors, we have been able to exchange ideas on various ways to promote this community of contributors to make Kiswahili a popular language in East Africa as well as one of the languages that computers can be taught to understand.
Can you tell us where the project has reached so far? What is going well and what can be improved?
This project has three phases. The first phase takes place on the Pontoon page. The main function of this page is to receive donations that help translate the Common Voice Kiswahili website, 86% of this work has been done and continues to be done. The second phase is the contribution of sentences that will be used to teach the machine to understand Kiswahili. Already 9,000 sentences have been compiled and approved and published on the page commonvoice.mozilla.org/sw. These sentences are used in the third phase which is the collection and verification of voices that will be used to train computers to understand Kiswahili. The project is currently in its third phase which is expected to be completed by the end of the year.
How do you think this project will contribute to the work of the entire Natural Language Processing (NLP) community and the Kiswahili language?
Often, African languages in this community are marginalized for lack of research resources. This task of localizing Kiswahili as well as teaching computers how ordinary people speak Kiswahili will be a great victory for this community as the Kiswahili language will be known by many in the world, thus contributing to more people learning the language.
In the next three months, where do you expect this project to have reached? What successes would you like to see in this project?
For me three months is a very short time, but I will say, I would like to see the Common Voice Kiswahili dataset launched. This will help researchers to start analyzing this dataset and also to start teaching computers to understand Kiswahili.
Is there anything you would like to add?
Many thanks to my colleague George Githuma and everyone in the community, for the support we received from our colleagues and the manager from Mozilla. It has been a pleasure to work with this great team and I hope that the future looks good for all of us.