GDC 2025 Session Viewer
Machine Learning Summit: Behind the Platform: Deploying ML for Voice Safety
Kiran Bhat (Senior Technical Director, Roblox)
Pass Type: All Access Pass, Summits Pass - Get your pass now!
Track: Programming
Format: Lecture
Vault Recording: TBD
Audience Level: Intermediate
For 3D immersive worlds and games, much like in the physical world, few things are more authentic or powerful than the human voice in forging lasting friendships and connections. But how do we scale the immersiveness and richness of voice communication while keeping our community safe and civil? In this session, we'll explore how Roblox brought to life Real-time Safety, an end-to-end machine learning (ML) model—operating at a scale of millions of minutes of voice activity per day—that detects policy violations in voice communication more accurately than human moderation. We'll then dive into how the outputs from this system feed into another model to determine appropriate consequences, and the triggers to notify people who have violated our policies, initially with warnings and then with more drastic actions if the behavior persists. One of our real-time voice safety models is now open source, so attendees will walk away armed with not only the learnings to make their in-game communication safer, but also the technology to get started today.
Takeaway
Learn how to scale voice communication while keeping gaming communities safe by leveraging novel machine learning techniques.
Intended Audience
Game developers, community managers, trust and safety teams, researchers, platforms, and AI/ML enthusiasts.
