Automatic VHF airband scanner and voice recorder

Scans the aviation band with an SDR, records every voice call on its own,
and lets you listen, browse and download them from a web browser.
Also decodes satellite pictures: SSTV from the ISS and CubeSats, and Meteor-M LRPT weather images.

Features · Supported SDRs · Screenshots · Install · Usage · Satellites · How it works · API

Support the project

---

Features

• 📡 Automatic scanning: hops across your frequency ranges (118–137 MHz by default) and stays on a range while someone is talking.
• 🎙️ Records every call: each transmission is saved on its own, with a short pre-roll so the first word isn't cut off.
• 🗣️ Voice detection: a built-in speech detector separates real voice from data bursts, dead carriers, tones and noise.
• 🎧 Play in the browser: click a waveform to seek, or turn on auto-play for new calls.
• 💾 MP3 and WAV: voice calls are saved as MP3 automatically; any recording can be downloaded as MP3 or WAV.
• 🌈 Spectrum page: live spectrum, waterfall and activity history, plus a spectrogram of every signal that was not recognised as voice.
• ✈️ Voice only: ACARS and VDL2 data channels are ignored by default.
• 📻 AM or NFM: AM for the airband, or narrow-band FM (marine, PMR, amateur, business radio), with 8.33 / 25 / 12.5 / 6.25 / 5 kHz channel rasters.
• 🛰️ Satellite images: SSTV pictures from the ISS (ARISS events, PD120 / PD180) and CubeSats (Robot 36 / 72, Martin, Scottie, all PD modes), and Meteor-M N2-3 / N2-4 LRPT weather images (false-colour composite plus every MSU-MR channel), decoded live and kept in a gallery. Doppler is followed automatically, so no tracking software is needed.
• ⚙️ Web settings: device, gain, ranges, thresholds and MP3 options, all from the browser.
• 🖥️ Single native program: C++17, no Python, runs as a desktop app or headless server on Windows and Linux; open it from any phone or PC on your network.

Supported SDRs

SDR	Models	Sample rates	Build requirement
RTL-SDR	Any RTL2832U dongle: RTL-SDR Blog V3 / V4, Nooelec NESDR (Mini, SMArt), generic R820T / R820T2 / R828D / E4000 sticks	1.024 – 3.2 MS/s (2.4 recommended)	librtlsdr (always required)
RTL-SDR over the network	Any rtl_tcp server: rtl_tcp on a Raspberry Pi / PC, SDR++ / SDR# rtl_tcp server, Android "RTL-SDR driver"	1.024 – 3.2 MS/s (2.4 recommended; LAN bandwidth ≈ 2 bytes per sample)	none (built in)
Airspy	Airspy R2, Airspy Mini	2.5 · 3 · 6 · 10 MS/s	libairspy (optional)
Airspy HF+	Airspy HF+ Dual Port, HF+ Discovery (VHF input: 60 – 260 MHz)	192 – 912 kS/s (768 recommended)	libairspyhf (optional)
SDRplay	RSP1, RSP1A, RSP1B, RSP2, RSPduo, RSPdx, RSPdx R2	2 – 10 MS/s	SDRplay API 3.x (optional)
HackRF	HackRF One	2 – 10 MS/s	libhackrf (optional)
PlutoSDR	ADALM-Pluto (Rev B/C/D), Pluto+ and other AD9361/AD9363 boards over USB or Ethernet	2.4 – 10 MS/s (2.4 recommended)	libiio 0.x or 1.x (optional)
IQ WAV file	8- or 16-bit stereo I/Q recordings (e.g. from SDR#, SDR++, SDRuno)	from the file	none (built in)

• Optional drivers are detected automatically at build time; only the SDRs that were found appear under Settings → Device type.
• A wider sample rate covers more of the airband at once (fewer hops), but uses more CPU.
• RTL-SDR Blog V4 needs the rtl-sdr-blog driver, a newer librtlsdr than most Linux distributions ship.
• Airspy HF+ has a narrow bandwidth (under 1 MHz), so it hops through 118–137 MHz in about 30 steps. For quicker coverage, narrow the scan ranges to your local frequencies. Gain -1 turns on AGC; 0–48 sets the attenuator (48 = none).

RTL-SDR over the network (rtl_tcp)

Put the dongle next to the antenna and run AirBandScan on another computer.

1. On the computer with the dongle, start the server so it listens on the network:

   rtl_tcp -a 0.0.0.0          # Debian / Raspberry Pi OS: sudo apt install rtl-sdr

2. In AirBandScan, choose Settings → Device type → RTL-SDR (rtl_tcp) and enter the server as host:port, e.g. 192.168.1.20:1234 (port 1234 is the default). Test checks the connection.

Gain, ppm and bias-T are sent to the server. A 2.4 MS/s stream is about 4.8 MB/s, so wired Ethernet or good Wi-Fi is needed. rtl_tcp has no security: only use it on a trusted network.

PlutoSDR

• Connection: set Settings → Pluto address.
  • USB: leave it empty, or use ip:192.168.2.1; a Pluto on USB also shows up as a network card at that address.
  • Ethernet (Pluto+, or a Pluto with a USB-Ethernet adapter): ip:<address>, e.g. ip:192.168.1.50 or ip:pluto.local.
  • A specific USB device: usb:1.4.5. Detect lists the Plutos found by USB scan and mDNS.
• Airband frequencies: a stock Pluto (AD9363) only tunes 325 – 3800 MHz. Enable the AD9364 range (70 MHz – 6 GHz) once:

  ssh root@192.168.2.1          # password: analog
  fw_setenv attr_name compatible
  fw_setenv attr_val ad9364
  reboot

  Without this, AirBandScan reports that the Pluto rejected the frequency.
• Gain: -1 = AGC (slow attack); 0 – 71 dB = manual.
• Sample rate: 2.4 MS/s is the safe choice over USB 2.0 and the USB network link; higher rates may drop samples.

Screenshots

Voice recordings

Spectrum & undecoded signals

Satellites (SSTV & Meteor-M LRPT)

Settings

Install

Download (prebuilt)

Ready-to-run packages are on the Releases page:

System	File
Windows 64-bit / 32-bit	airbandscan-*-windows-x64.zip / -windows-x86.zip
Linux 64-bit / 32-bit	airbandscan-*-linux-x86_64.tar.gz / -linux-i386.tar.gz
Raspberry Pi 64-bit / 32-bit OS	airbandscan-*-raspberrypi-arm64.tar.gz / -raspberrypi-armhf.tar.gz

Linux and Raspberry Pi builds need Debian 12, Ubuntu 24.04, Raspberry Pi OS (bookworm) or newer, plus sudo apt install librtlsdr0 libairspy0 libairspyhf1 libhackrf0 libiio0 libmp3lame0. Releases are built by GitHub Actions whenever a version tag such as v1.0.0 is pushed.

Build from source

Linux (Debian / Ubuntu)

sudo apt install build-essential cmake ninja-build pkg-config librtlsdr-dev libmp3lame0
# optional: libairspy-dev libairspyhf-dev libhackrf-dev libiio-dev   (SDRplay: install the SDRplay API 3.x)

cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release
ninja -C build
./build/airbandscan

Fedora: rtl-sdr-devel lame-libs · Arch: rtl-sdr lame

Windows (MSYS2)

Open the MSYS2 UCRT64 shell:

pacman -S --needed mingw-w64-ucrt-x86_64-toolchain mingw-w64-ucrt-x86_64-cmake mingw-w64-ucrt-x86_64-ninja \
  mingw-w64-ucrt-x86_64-pkgconf mingw-w64-ucrt-x86_64-rtl-sdr mingw-w64-ucrt-x86_64-lame
# optional: mingw-w64-ucrt-x86_64-hackrf mingw-w64-ucrt-x86_64-libiio
# Airspy HF+: MSYS2 has no package; see the libairspyhf step in .github/workflows/release.yml

cmake -S . -B build -G Ninja
ninja -C build
./build/airbandscan.exe

32-bit Windows: MSYS2 no longer packages librtlsdr for 32-bit, so build it from source; see the windows-x86 job in .github/workflows/release.yml.

The needed DLLs are copied next to airbandscan.exe, so it also starts with a double-click. For an RTL-SDR on Windows, install the WinUSB driver with Zadig.

MP3: LAME is loaded at run time (libmp3lame.so.0 / libmp3lame-0.dll). If it is missing, a lame or ffmpeg command on PATH is used instead. Without either, WAV still works and the Settings page tells you MP3 is unavailable.

Usage

airbandscan [-c airband.conf] [-p 8080] [-d recordings] [--no-autostart]
airbandscan --decode-sstv FILE | --decode-lrpt FILE [--meteor m2x|m2] [-o DIR]

1. Plug in your SDR and start airbandscan.
2. Open http://localhost:8080/, or http://<computer-ip>:8080/ from a phone or tablet.
3. Under Settings, choose your device, set the gain (about 30–40 dB on an RTL-SDR) and enter your local frequencies.
4. Calls appear on Voice recordings as they happen.

On first start a commented airband.conf is created; see airband.conf.example.

Tips

Tip	Why
List only the frequencies used near you, e.g. 118.1-119.0, 121.5-122.0	One RTL-SDR sees about 1.9 MHz at a time; 118–137 MHz needs 10 hops, and calls on another range are missed while one is being recorded
Use a fixed gain rather than AGC	AGC makes the noise floor move
Too much noise saved as voice? Raise Voice threshold	Lower it if weak calls land on the Spectrum page
Don't want ATIS / VOLMET? Add them to Ignore	Continuous broadcasts are otherwise split every Maximum length seconds
Scanning FM radio (marine, PMR446, amateur)? Set Modulation to NFM, Channel width 12.5 kHz (25 kHz channels: 16 kHz) and the matching Channel raster, e.g. 6.25 or 12.5 kHz	Airband voice is AM; most other VHF/UHF voice is narrow FM. All scan ranges use the same modulation

Satellite images (SSTV and Meteor LRPT)

Open Satellites, choose a satellite and press Start receiving. The SDR then stays on that frequency instead of scanning the airband. Stop & resume scanner goes back to scanning. The mode you used last is the one started at launch.

Preset	Frequency	What you get
ISS — SSTV	145.800 MHz FM	ARISS SSTV events, usually PD120 or PD180 (640 × 496)
Meteor-M N2-3 / N2-4 — LRPT	137.900 MHz (137.100 MHz alternate)	1568 px wide MSU-MR images: 221 false-colour composite + each channel
Other	any	SSTV from CubeSats or amateur stations (e.g. 435–438 MHz), or LRPT on another frequency

• Leave it running: every SSTV picture is found by its VIS header and saved on its own. A Meteor pass is saved when its signal has been gone for 90 s. Pictures appear in Decoded images and update live while they arrive.
• No tracking needed: Doppler (±3.5 kHz) and dongle frequency error (up to ±20 kHz for LRPT) are followed automatically.
• SSTV modes: PD50, PD90, PD120, PD160, PD180, PD240, PD290, Robot 36, Robot 72, Martin M1 / M2, Scottie S1 / S2 / DX. Slant from the transmitter's clock error is corrected from the sync pulses.
• Meteor LRPT: 72 ksym/s OQPSK with NRZ-M (N2-3 and N2-4); choose N2 (QPSK) for recordings of the retired Meteor-M N2. The 80k interleaved test mode is not supported.
• Antenna: a V-dipole or QFH tuned for 137 MHz for Meteor; a vertical or small Yagi for the ISS. LRPT needs a clear sky view; SSTV from the ISS works with a simple antenna.
• SDR: any of the supported SDRs. An RTL-SDR runs at 1.024 MS/s in satellite mode to save CPU; other SDRs use the sample rate from Settings. PlutoSDR needs the AD9364 range for 137 / 145 MHz.
• Orientation: on northbound passes the Meteor image is upside down (north at the bottom). That is how the satellite scans.

Find pass times with any satellite tracker (Heavens-Above, Gpredict, the ARISS SSTV event page).

Decode a recording from the command line

# SSTV from receiver audio (any WAV: from a radio, SDR# / SDR++ audio recording, …)
airbandscan --decode-sstv iss_pass.wav -o pictures

# Meteor LRPT from an IQ recording centred on the signal (≥ 144 kS/s), 8-bit soft symbols (.s) or CADUs (.cadu)
airbandscan --decode-lrpt meteor_baseband.wav -o pictures
airbandscan --decode-lrpt old_meteor_m2.wav --meteor m2 -o pictures   # retired Meteor-M N2 (QPSK)

How it works

 SDR ──► range hopping ──► FFT + noise learning ──► detection ──► per-call channel ──► AM / NFM demod ──► voice check
                                                                                                      │
                                       Voice recordings (WAV + MP3)  ◄── voice ───────────────────────┤
                                       Spectrum & undecoded (+ spectrogram) ◄── not voice ────────────┘

Satellites (one SDR, instead of the scanner):

 SDR ──► channel ──► FM demod ──► 1900 Hz FM-to-frequency ──► VIS header ──► sync-locked lines ──► SSTV picture
     └─► channel ──► (O)QPSK demod (x⁴ FFT acquisition, Costas, Gardner) ──► Viterbi r=½ K=7 ──► NRZ-M
         ──► 1ACFFC1D frame sync ──► derandomise ──► Reed-Solomon (255,223)×4 ──► CCSDS packets ──► MSU-MR JPEG ──► image

1. Scan: ranges are split into windows of 80 % of the sample rate; the scanner hops between them.
2. Detect: every 20 ms a Welch-averaged FFT is compared with each window's learned noise floor. A signal start_level dB above noise opens a call, snapped to the 8.33 kHz airband raster; it closes after timeout_ms below stop_level.
3. Record: each call gets its own channelizer and AM or NFM demodulator (8 kHz mono audio), seeded with a pre-roll so nothing is clipped, and a squelch gate removes the hiss before and after the carrier.
4. Classify: pitch periodicity, pitch movement, loudness variation and modulation depth decide voice or not voice. You can move any recording between the two lists yourself.

Recordings on disk

recordings/2026-10-14/20261014-131915-832-0119700.wav    audio (8 kHz mono)
                                               ….mp3    MP3
                                               ….spg    spectrogram
                                               ….meta   time, frequency, level, verdict

The list is rebuilt from the .meta files at start-up, so recordings can be copied, moved or deleted with normal file tools.

Satellite images go to recordings/images/YYYY-MM-DD/: <id>.png (picture or LRPT composite), <id>_thumb.png, <id>_ch1.png … (LRPT channels) and <id>.imeta.

HTTP API

Method	Path	Description
GET	/api/status	scanner state, live calls, counts
GET	/api/spectrum	live spectrum and noise floor
GET	/api/history?w=N	activity waterfall (binary)
GET	/api/recordings?kind=voice|other|all&limit&offset&freq=MHz	list recordings
GET	/rec/<id>.wav · .mp3 · .spg (?dl=1 to download)	recording files
POST	/api/rec/<id>/voice?v=1|0 · /api/rec/<id>/delete	relabel / delete
POST	/api/recordings/delete?kind=other	delete a whole list
GET / POST	/api/config	read / save settings
POST	/api/start · /api/stop	start / stop (the scanner, or the satellite receiver in satellite mode)
POST	/api/sat/start (sat_mode=sstv|lrpt, sat_freq_mhz, sat_lrpt_variant=m2x|m2) · /api/sat/stop	satellite receiver on / back to the scanner
GET	/api/sat/status · /api/sat/live.png	receiver state, lock, constellation · picture in progress
GET	/api/images?kind=sstv|lrpt&limit&offset · /img/<id>.png (_thumb, _chN)	decoded images
POST	/api/img/<id>/delete	delete an image

⚠️ The web interface has no login. Keep it on your home network, or set Listen address to 127.0.0.1.

Testing without an SDR

# Make a test IQ file (add --fm for narrow-band FM): voice from speech files + data tone, carrier, noise, short burst
python3 tools/make_test_iq.py test_iq.wav speech1.wav speech2.wav
# Settings → Device type "IQ WAV file", file test_iq.wav, centre 120 MHz

# Range-hopping test with a simulated SDR
ninja -C build hop_test && ./build/hop_test

# SSTV: a test picture in any mode, as audio, or FM-modulated IQ with Doppler and noise
python3 tools/make_test_sstv.py PD120 sstv.wav && ./build/airbandscan --decode-sstv sstv.wav
python3 tools/make_test_sstv.py PD180 iss_iq.wav --iq --rate 250000 --offset 40000
# → device "IQ WAV file", centre 145.76 MHz; Satellites → ISS

# Meteor LRPT: re-modulate decoded CADUs (e.g. a SatDump .cadu) as OQPSK with Doppler and noise
python3 tools/make_test_lrpt.py pass.cadu lrpt_iq.wav --snr 6 && ./build/airbandscan --decode-lrpt lrpt_iq.wav

Credits

AirBandScan builds on three open-source projects:

• AirScope: SDR drivers and DSP (voice path only; ACARS, VDL2 and ADS-B removed)
• rtl-sdr-scanner-cpp: scanning and transmission-detection logic, ported off GNU Radio
• sdr-monitor: the idea for the recordings and spectrogram web pages

The SSTV and LRPT decoders are written for AirBandScan. The Meteor link parameters and MSU-MR image format follow SatDump and meteor_decoder. PNG output uses stb_image_write (public domain).

License

GPLv3, the same as the projects it is based on.