Notes on voice technology and real-time protocols

Voice on a network is a chain of compromises: sample the air, compress it, packetise it, ship it, jitter-buffer it, decompress it, and play it back before the talker notices.

These notes are about the links in that chain.

• G.711 is not dead, and that is fineSixty-four kilobits of uncompressed speech still carries most of the world's phone calls.
• Why Opus became the defaultIt is not the best codec at any single bitrate. It is the only one that is good across all of them.
• SIP and XMPP solve different problemsOne is a session protocol, the other is a presence and messaging protocol that learned to carry voice.
• WebRTC signalling is deliberately not in the specThe standard defines how two endpoints talk. It says nothing about how they find each other, and that is the whole difficulty.