How Norvane delivers low-latency voice AI at scale

A split relay and transceiver architecture for WebRTC at scale.