5 ms·
Congratulations on the launch! I'm curious to understand, in your experience, what was the most challenging part of building a realtime voice AI app was? Naivel
by shravan 2y ago
Congratulations on the launch! I'm curious to understand, in your experience, what was the most challenging part of building a realtime voice AI app was? Naively, assumed that this would be a solved problem.
- jankboy 2y agoThe hardest part has to be building webRTC streaming support from scratch into our SDK to provide extremely low latency API. We had this idea where we imagined the webRTC server and AI model running on the same instance and talking to the user via a direct peer connection. As far as we know, this architecture is first of its kind and isn't usually how webRTC peer connections are made.