tr-plugin-dvcf

Trunk Recorder plugin that captures raw IMBE/AMBE voice codec frames and saves them as .dvcf files and/or publishes them over MQTT.

This is the companion plugin to IMBE-ASR — it provides the raw codec data that IMBE-ASR uses to transcribe P25 radio calls without audio reconstruction.

Why

Standard ASR pipelines for P25 radio work like this:

codec params → reconstruct audio (libimbe) → mel spectrogram → ASR

Every step is lossy. IMBE-ASR skips all of it by reading the codec parameters directly. But to do that, you need the pre-vocoder codec frames — which trunk-recorder normally discards after audio synthesis.

This plugin taps trunk-recorder's voice_codec_data() callback to capture the raw frames before they're consumed by the vocoder.

Features

• File writing — saves .dvcf sidecar files alongside audio recordings (same base name, .dvcf extension)
• MQTT publishing — publishes codec frames as audio_dvcf_base64 in the MQTT call-end message, compatible with tr-engine
• Analog-safe — only digital calls (P25, DMR, D-STAR, YSF) produce .dvcf files; analog calls are ignored automatically
• Stale-call reaper — detects calls that never receive a call_end (a known trunk-recorder bug) and salvages their data instead of leaking resources
• Both features independently toggleable via config

File Format

.dvcf = Digital Voice Codec Frames. SymbolStream v2 binary format — see DVCF_SPEC.md for the full specification.

Files are a concatenation of binary messages (8-byte SSSP headers + codec frame payloads) representing a single call. They can be:

• Played back for offline transcription
• Streamed directly to a TCP consumer
• Uploaded to the IMBE-ASR server via HTTP

Requirements

• Trunk Recorder with voice_codec_data() plugin API (merged in TrunkRecorder/trunk-recorder#1098)
• libmosquitto / Paho MQTT C++ (for MQTT publishing)
• Boost (already required by trunk-recorder)

Building

Builds as a user_plugins drop-in — no fork of trunk-recorder required. Requires trunk-recorder with voice_codec_data() plugin API support (v5.0+).

# 1. Clone trunk-recorder
git clone https://github.com/TrunkRecorder/trunk-recorder.git
cd trunk-recorder

# 2. Drop this plugin into user_plugins/
mkdir -p user_plugins
git clone https://github.com/trunk-reporter/tr-plugin-dvcf user_plugins/mqtt_dvcf

# 3. Build with local plugins enabled
cmake -B build -DUSE_LOCAL_PLUGINS=ON
cmake --build build -j$(nproc)

# 4. Install
sudo cmake --install build

The compiled libmqtt_dvcf.so will be installed to /usr/local/lib/trunk-recorder/ (or your configured prefix).

Dependencies

• Paho MQTT C++ library (libpaho-mqttpp3-dev)
• Paho MQTT C library (libpaho-mqtt3as-dev)
• Boost (already required by trunk-recorder)

# Ubuntu/Debian
sudo apt-get install libpaho-mqtt3as-dev libpaho-mqttpp3-dev

Configuration

Add to your trunk-recorder config.json:

{
  "plugins": [
    {
      "name": "mqtt_dvcf",
      "library": "libmqtt_dvcf",
      "write_enabled": true,
      "mqtt_enabled": true,
      "broker": "tcp://localhost:1883",
      "topic": "tr/feeds",
      "clientid": "dvcf-plugin",
      "username": "",
      "password": "",
      "qos": 0,
      "stale_call_timeout_sec": 300
    }
  ]
}

Option	Default	Description
write_enabled	true	Write .dvcf sidecar files to disk
mqtt_enabled	false	Publish codec frames over MQTT
broker	tcp://localhost:1883	MQTT broker URL
topic	trunk-recorder	MQTT topic prefix (publishes to {topic}/dvcf)
clientid	dvcf-handler	MQTT client ID
username	""	MQTT username (optional)
password	""	MQTT password (optional)
qos	0	MQTT QoS level
stale_call_timeout_sec	300	Seconds of inactivity before a call without call_end is reaped. Data is salvaged to .stale files when possible.

MQTT Message Format

When mqtt_enabled: true, the plugin publishes a JSON message on {topic}/dvcf containing the base64-encoded .dvcf content and call metadata:

{
  "audio_dvcf_base64": "<base64-encoded .dvcf content>",
  "metadata": {
    "talkgroup": 9170,
    "talkgroup_tag": "Fire Dispatch",
    "talkgroup_alpha_tag": "BUTCO FD",
    "talkgroup_group": "Fire",
    "freq": 855737500,
    "start_time": 1711234567,
    "stop_time": 1711234590,
    "call_length": 23,
    "signal": -42.5,
    "noise": -110.2,
    "freq_error": 12,
    "spike_count": 0,
    "emergency": false,
    "priority": 3,
    "phase2_tdma": false,
    "tdma_slot": 0,
    "short_name": "butco",
    "filename": "9170-1711234567_855737500.wav",
    "srcList": [{"src": 1234567, "time": 1711234567, "pos": 0.0, "emergency": 0, "signal_system": "", "tag": "Engine 5"}]
  }
}

tr-engine recognizes the audio_dvcf_base64 field and routes the call to the IMBE-ASR transcription provider.

Integration

trunk-recorder + mqtt_dvcf plugin
    → MQTT (audio_dvcf_base64)
    → tr-engine (STT_PROVIDER=imbe)
    → imbe-asr server
    → transcript stored in database

For live streaming (lower latency), see symbolstream which streams frames in real time over TCP/UDP.

Related Projects

• IMBE-ASR — ASR model that reads .dvcf files directly
• tr-engine — backend that ingests MQTT and calls IMBE-ASR
• symbolstream — real-time codec frame streaming plugin