The Audio Layer 3.0: Voice x Robotics
Voice is becoming one of the most powerful ways we interact with AI.
Real-time agents and synthetic voices only work when they're built to withstand the noise, variability, and unpredictability of real-world environments, not just the safety of controlled tests. That same challenge is now showing up wherever AI meets the physical world - including robotics, where perception has to hold up just as well outside the lab.
That's why on September 15, we're hosting The Audio Layer 3.0 at Tavus's office in San Francisco - the next edition of our gathering for people shaping the future of Voice AI, this time widening the lens to where voice and robotics intersect. Together, we'll explore how the audio and perception foundation of modern agents is evolving, and what it takes to make these systems reliable, observable, and resilient when the world behaves unpredictably.
Program (tentative, pending final confirmation)
6:00 PM — Doors open, networking
6:30 PM — Live demos
7:00 PM — Panel discussion
7:30 PM — Networking continues
On the panel:
Tavus — Quinn Favret, Co-founder
A "Human Computing" AI lab building real-time Conversational Video Interfaces — full-duplex face rendering, multimodal perception, and turn-taking, all running sub-500ms. Teasing their new Kiosk mode at the event.ai-coustics — Fabian Seipel, Co-Founder
Building the audio intelligence layer for voice AI - primary speaker isolation, audio insight, and voice activity detection models that keep STT and turn-taking systems reliable on real customer calls.LiveKit — David Chen, GM Robotics
An open-source framework and cloud platform for building, deploying, and scaling real-time AI agents across voice, video, and robotics. Powers production conversational agents from speech recognition through telephony and global deployment.Gradium — Constance Grisoni, Chief Growth Officer
Advanced voice AI for natural-sounding text-to-speech, transcription, voice cloning, and real-time translation - built for developers and enterprises shipping conversational AI.Lightberry — Ali Attar, Founder
Makers of Lumi, billed as "the world's most interactive robot" - software and robotics that let machines communicate naturally, read their environment and the people in it, and improve through interaction.
What to expect
Live demos from the robotics and voice AI builders in the room - including Lightberry's robot Lumi - followed by a panel discussion across the voice and robotics stack. Plus a chance to connect with fellow builders over food and drinks.
Open to anyone working in Voice AI or robotics - real-time audio engineers, agent and robotics builders, infrastructure and research teams alike.
Space is limited - RSVP early.
We look forward to seeing you there!