Media Streams

Live call audio over a WebSocket, as JSON messages: start one with Connect><Stream> (two-way) or Start><Stream> (one-way).

The connection

We open your wss:// URL with X-DWW-Signature on the upgrade (see Webhooks), and check your certificate for the hostname.

Messages to you

• connected, then start (with streamSid, callSid, your customParameters, and mediaFormat audio/x-mulaw, 8000 Hz, 1 channel);
• media: 20 ms of base64 µ-law each;
• dtmf, when the caller presses a key;
• mark, echoed when the audio before it has actually played;
• stop.

Messages from you (two-way streams)

media (µ-law 8 kHz), mark and clear.

Limits

• a message up to 64 KiB, and a payload up to 1 second of audio;
• up to 100 messages a second, bursts of 200;
• we ping every 10 seconds and close a stream with no pong for 30 seconds (730003); your silence is fine;
• an invalid message closes the stream (730002).

Timing

start is sent once audio flows, about 1.4 seconds after the answer. Audio you send before start is dropped.