the project includes a websocket endpoint for realtime decoding. will add it to the docs.

we are already using it for a callcenter with around 50 parallel audio streams.