AccessDR

Note: This project is a proof of concept in a very early alpha stage. Expect rough edges, missing features, and breaking changes.

Documentation | Download

Accessible SDR radio application designed for blind and visually impaired users. Screen reader driven, fully keyboard operated, with spectrum sonification. Built with Python, wxPython, and pyrtlsdr.

Development Setup

Requires Windows 10/11, Python 3.12+, and an RTL-SDR dongle with WinUSB driver (Zadig).

git clone https://github.com/ragb/accessdr.git
cd accessdr
python -m venv .venv
.venv\Scripts\activate
pip install -e ".[dev]"
invoke fetch-dlls
python main.py

invoke fetch-dlls downloads the librtlsdr static build. Alternatively, copy rtlsdr.dll from librtlsdr releases.

Architecture

core/
  sdr_device.py          # RTL-SDR abstraction + IQ capture thread
  audio.py               # WASAPI audio output via sounddevice
  dsp/
    demodulator.py       # Stateful WFM/NFM/AM/SSB/CW demodulators
    filters.py           # Decimation and filtering
    mixer.py             # Software VFO offset mixer
    noise_blanker.py     # Impulse noise blanker (median + interpolation)
    ctcss.py             # CTCSS tone detector (generalized Goertzel)
    rds.py               # RDS/RBDS decoder (PS, RT, PTY)
    spectrum.py          # FFT spectrum analyser + peak detection
  scanner.py             # Frequency scanner
accessibility/
  speech.py              # Screen reader output (accessible_output2)
  sonification.py        # Spectrum-to-audio tone mapping + probe tone
config/
  settings.py            # JSON-persisted settings
  bookmarks.py           # Bookmark storage
  bands.py               # Band definitions
ui/
  main_window.py         # Primary window + keyboard shortcuts
  spectrum_panel.py      # Line graph + waterfall spectrogram display
  colormaps.py           # CVD-safe colormap LUTs (Viridis, Magma, Grayscale)
  dialogs/               # RF, audio, spectrum, scanner, bookmarks, help
locale/                  # Translations (gettext .po/.mo)
installer/               # NSIS installer script

DSP Pipeline

RTL-SDR @ 2.4 MSPS
  → noise blanker (impulse suppression)
  → decimate ×10 → 240 kSPS baseband
  → software mixer (VFO offset)
  → demodulate (stateful filters)
      WFM: stereo blend + RDS decode
      NFM: CTCSS tone detect
  → resample_poly → 48 kHz stereo
  → sounddevice WASAPI output

Cross-thread communication uses queue.Queue + wx.CallAfter. Live tuning goes through raw librtlsdr C calls for thread safety while read_samples blocks.

Building

invoke fetch-dlls    # download RTL-SDR DLLs
invoke build         # DLLs + translations + PyInstaller freeze
invoke installer     # full build + NSIS installer
invoke clean         # remove build artifacts

NSIS required for invoke installer (download).

Internationalisation

Uses gettext. Portuguese (pt_PT) included.

invoke extract-messages   # update locale/accessdr.pot
# edit locale/pt_PT/LC_MESSAGES/accessdr.po
invoke compile-messages   # build .mo files

License

MIT