Idioma: English (mozilla)
Technologists and curious practitioners are welcomed to hear a talk about how governance considerations can be baked into every stage of building a speech recognition algorith - from data collection through to model testing. There will be several interactive scenarios for workshopping together.
In four expert-guided subsections, we'll be looking at the life cycle of a speech AI project from a governance and design perspective.
First, we'll learn about data design and the data considerations you need to take before you begin implementing your project plan.
From data design, we'll look at community mobilization. How to get relevant language and technology community involvement in your speech AI project.
We'll next examine how to recognize, control for and reduce bias in your voice AI datasets and models.
Finally, we'll examine issues of inclusive uptake and community mobilization. How to grow an ecosystem of use and contributions around your new project, feeding back into data design for your next cycle of iteration.
Common Voice Product Lead | Mozilla Foundation
EM Lewis-Jong oversees platform strategy and direction for Mozilla Common Voice. She has a decade of experience in product, tech policy, and applied research, most recently in the GovTech space, where she ran product teams from concept phase to Series A. She started her career as an investigative analyst at a Global Fund, and advocated for legislative amendments with peers at the UK House of Lords. Nominated for the Women in Tech and Rising Stars of Tech awards, she has spoken at FWD50 and Republica, and holds a Churchill Fellowship for her fieldwork on citizen engagement technologies. She studied at the University of Oxford and is a SheCodes Alumna.
Technical community manager, Common Voice