simplestream for Trunk Recorder Pro

Streams the audio of calls as Trunk Recorder Pro records them to other programs, over UDP or TCP. It does what Trunk Recorder's simplestream plugin does, in the same packet formats, so programs written for that work with this.

The audio is raw PCM: 16-bit, mono, little-endian, 8 kHz (analog channels too). That's about 128 kbit/s per call, which is fine on one computer or a home network but not meant for the internet: the programs receiving it might play it, mix it, or compress it and stream it on.

Settings

A list of streams. Each sends some of the audio to one place:

Setting
Send to	udp://host:port or tcp://host:port.
Talkgroup	The talkgroup to stream. Calls patched with it count too. 0 streams every talkgroup.
System	The short name of the system to stream. Leave it empty for every system.
Talkgroup in each packet	Put the talkgroup before the audio in each packet.
JSON in each packet	Put the call's details before the audio in each packet, instead of the talkgroup.
Packets when calls start	With JSON: also send a packet of JSON, with no audio, when a call starts.
Packets when calls end	With JSON: also send one when a call ends.

Talkgroup numbers repeat between systems, so stream each system to a port of its own, or turn on JSON (which names the system).

Packets

Each packet is one slice of audio, with, depending on the settings:

• Nothing before it.

• The talkgroup: 4 bytes, little-endian, then the audio.

• JSON: its length in 4 bytes, little-endian, then the JSON, then the audio:

  {"audio_sample_rate":8000,"event":"audio","freq":851012500,"patched_talkgroups":[101],
   "short_name":"county","src":1234,"src_tag":"","talkgroup":101}

  A call start is the same with "event":"call_start" (plus talkgroup_tag and patched_talkgroup_tags) and no audio; a call end has talkgroup, patched_talkgroups, freq, short_name and "event":"call_end".

talkgroup in an audio packet is the talkgroup the stream is for, which is the patched talkgroup when the call is on another. src is the last radio heard when the call started (-1 when none was).

examples/example_audio_player.py (from Trunk Recorder) plays a UDP stream. To play one through PulseAudio on Linux, load its TCP module and stream to it:

pacmd load-module module-simple-protocol-tcp sink=1 playback=true port=9125 format=s16le rate=8000 channels=1

with Send to tcp://127.0.0.1:9125.

What it does when it can't send

Audio that can't be sent right away is dropped: it's live. A TCP stream connects when its first audio comes, and connects again (every few seconds) after the connection breaks; until then the plugin shows a warning.

Differences from Trunk Recorder

• With Talkgroup 0, a patched call is sent once, not once for each talkgroup in the patch.
• src_tag is always empty.
• Calls end when the recorder ends them, a little before their files are written (where Trunk Recorder sent call_end).

Coming from Trunk Recorder

Trunk Recorder's streams can be pasted in as they are, including the older address, port and useTCP settings. The recorder sends live audio to plugins that ask for it: there's no audioStreaming setting to turn on.

Building

cargo build --release

License

GPL-3.0-or-later, like Trunk Recorder.