Mozilla Common Voice Fellows are inviting all Kiswahili speakers, voice technologists, and data scientists to participate in the Kiswahili Festival slated for February 24 to 25.
This two-day event, hosted in partnership with Swahilipot Hub in Mombasa, Kenya, aims to bring together community members to help grow the Common Voice dataset for the Kiswahili language.
Common Voice is a multilingual crowdsourced dataset powered by volunteer contributors worldwide, and it enables the development of inclusive machine learning models for voice applications.
The dataset allows technologists to train machine learning models to build voice applications.
By participating in the festival, contributors can help grow the dataset and increase awareness of Common Voice.
On the first day, participants will learn about Common Voice and how to contribute to the project. They will also have the opportunity to validate other contributed voices, and there will be prizes for the top contributors.
The second day will feature a coding challenge open to everyone across Kenya, where participants will have the chance to train and test a speech-to-text model on Mozilla’s Common Voice Kiswahili data.
The top projects will be awarded a total cash prize of KES 100,000 during the event.
Mozilla Fellow and event host Britone Mwasaru says, “This event is part of a community approach towards building the open voice dataset for the Kiswahili language on the Common Voice platform. It’s about lowering barriers to building and reducing bias in tools/products created. But it’s also about developing a language dataset by and for us. And I am very excited to see what that enables.”