radiotelegram

Telegram voice messages <> Baofeng radio RX/TX two-way bridge (half-duplex) with advanced audio processing.

description

radiotelegram is a software-hardware project allowing you to to communicate with peers in Telegram chat using your handheld radio (walkie-talkie).

• RX is recorded and sent to Telegram chat (topic) as a voice message.
• incoming chat voice messages are played into radio TX with enhanced audio processing.

enhanced audio processing features

• Advanced Squelch: Multi-criteria signal detection with spectral analysis and adaptive noise floor
• Voice Detection: Frequency domain analysis optimized for voice communications (300Hz-3.4kHz)
• Radio-optimized Processing: Pre-emphasis, bandpass filtering, and dynamic range compression
• Noise Reduction: Spectral noise reduction with adaptive thresholds
• PTT Reliability: Wake tone generation for consistent radio keying
• Quality Enhancement: Loudness normalization and overmodulation prevention

hardware

Hardware requirements:

• UR-5V Baofeng handheld radio (or compatible)
• PC audio adapter
  • buy APRS K1 PRO
  • DIY adapter (see schematics below)

adapter DIY guide

Components:

• (usb) audio card
• 2.5mm jack cable
• 3.5mm jack cable
• ethernet transformer — you can find one in unused electroncs: old routers, pc motherboards or other devices with ethernet ports.

Jack	Contact Points
3.5mm Baofeng	Sleeve, Ring
2.5mm Baofeng	Sleeve, Tip
3.5mm Microphone	Sleeve, Tip
3.5mm Headphones	Sleeve, Tip

1. find a datasheet for your ethernet transformer
2. select two pairs as shown on the schematics
3. cut cables, solder wires, that's it
4. polarity is not that crucial, for voice it's ok both ways.

config

1. copy .env.example to .env
2. edit file: fill bot token, group id and topic id (if present).

run with docker

docker compose up

raspberry pi

put these lines into /boot/firmware/config.txt

dtparam=audio=off
dtoverlay=vc4-kms-v3d,noaudio

to disable internal & hdmi audiocard

run manually

sudo apt install ffmpeg python3-venv python3-dev build-essential 

Note: pyaudioop was disconutinued in python3.13 so audioop-lts package is used.

uv sync
source ./venv/bin/activate
python3 ./radiotelegram/main.py

tuning (required for diy cards)

1. put baofeng into VOX mode (1/10).
2. adjust main volume knob so that audio doesn't clip (<100% volume).
3. adjust pc audio output volume so there is no voice distortion on the receiving end.

less is more. do NOT use full volume.

license

MIT

credits

thx to claude sonnet 4 for comprehensive audio analysis and filter code generation.