adswill/OnAir

• DVB-T2 / DVB-T (detected automatically; T2-Lite included): Europe, Middle East, Africa, Asia, Australia.
• ATSC 1.0 (North America, South Korea) and ATSC 3.0 NextGen TV (ROUTE, HEVC video with AAC audio, optional AC-4 audio through libavcodec when installed; needs 6.5 Msps or more; MMTP services are not supported).
• ISDB-T (Japan, Brazil and most of South America, Philippines; 13 segments, up to three layers, one-segment mobile TV).
• DVB-S / S2 / S2X satellite TV through an LNB, all MODCODs, VL-SNR frames, polarisation warning.
• TV sat beacon decoder (the narrow carrier of a geostationary beacon, found anywhere in the capture and tracked: level, C/N0, offset, drift; the reference carrier; phase-modulated telemetry recovered as bits and frames with the frame length learned from the repetition; CSV logging; a decoded view of every word of the frames with its confidence, the satellite estimated from the frequency, CSV and JSON export; details).
• DTMB (China, Hong Kong, Macau 8 MHz, Cuba 6 MHz).
• DVB-C cable (EN 300 429 / J.83 annex A and C, 16 to 256-QAM, symbol rate and QAM found automatically; needs a cable feed).
• Analog TV: PAL, SECAM, NTSC; systems B/G, I, D/K, M, N, H, K', 405-line A, C, E (819 lines), L, L'; NBTV and mechanical TV. Picture and sound.
• FPV video (analog): drone video links on the 5.8 GHz bands A, B, E, F, R (Raceband), L and the 1.2 GHz plan: every channel with its level, a scan that tunes to the strongest transmitter, the live picture (PAL or NTSC, frequency modulated) with its 6.0 / 6.5 MHz sound, snapshots and recording; the on-screen display is part of the picture, not decoded; needs 20 Msps (details).
• MAC: D2-MAC and D-MAC satellite TV (FM, 27 MHz; 20 Msps sees most of it); picture, channel name, digital sound; EuroCrypt is detected, not descrambled.

• DAB / DAB+ with a transmitter map from the TII, Journaline and MOT slide show.
• FM (87.5 - 108 MHz): stereo and RDS (station name, radio text, programme type, traffic flags), band scan, saved stations, Mono switch.
• AM stereo (C-QUAM) on medium wave: the 25 Hz pilot found, left and right from the phase of the carrier, a stereo / mono indicator; plain AM stations play as mono; the older Kahn-Hazeltine, Harris and Magnavox systems are not decoded (details).
• DRM (DRM30 and DRM+) on long, medium and short wave: audio, station information and the data services (slide show, website, EPG, TPEG, Journaline).
• AM broadcast (medium wave 531 - 1602 kHz and the shortwave broadcast bands, in the general receiver with MW and SW presets).
• HD Radio (NRSC-5, FM and AM) and CDR (China Digital Radio). HDC and DRA audio are patented or of unclear status: see Good to know. HD Radio and CDR sound play only with a codec add-on you build yourself (docs/CODECS.md); the stations, services and data work without it.

• ADS-B (1090 MHz): aircraft on a map with altitude, speed and callsign; aviation or metric units; feeds from other receivers; offline registrations and countries; range statistics.
• Mode A/C (1090 MHz secondary-radar replies): squawk and altitude of aircraft that answer radar, the ident pulse, emergency squawks, matched to the Mode S aircraft of the same capture on the ADS-B map; details.
• UAT (978 MHz, DO-282): ADS-B reports, FIS-B weather products, TIS-B traffic relay. Mode S uplink (1030 MHz, needs 8 Msps): interrogator codes, TCAS coordination.
• ACARS, VDL Mode 2 (every channel at once), VDL Mode 4 (STDMA; spec-based, no live captures validated yet) and HFDL (every channel in the band at once), with positions on a map.
• Airband (AM voice, several 25 and 8.33 kHz channels at once, scan, priority), VOR, ILS and NDB.
• FLARM (868.2 / 868.4 MHz): the old protocol is decoded; the newer one is only counted. Aircraft with the no-track or stealth flag are never shown, logged or exported.
• DJI drones (2.4 and 5.8 GHz, 16 Msps or more): OcuSync-like bursts are classified by shape (a guess, never an identification); DroneID is decoded. Drone Remote ID (ASTM F3411, BLE and Wi-Fi). Receive only.

• AIS ship positions, VDES (ASM decoded; VDE-TER only detected), NAVTEX, DSC and Weather fax.

• GNSS (GPS L1; satellites, position, time and a computer-clock check).
• Weather pictures: Meteor-M LRPT (137.1 / 137.9 MHz, Doppler tracking, channels and colour composite), GOES HRIT (1694.1 MHz), Elektro-L HRIT / LRIT and GGAK, HRPT / AHRPT (MetOp, FY-3, retired NOAA). The L-band ones need a dish with an LNA at the feed.
• Inmarsat-C, Inmarsat Aero, Iridium (bursts can be saved for iridium-toolkit), Orbcomm, Globalstar (activity monitor only; L-band 1610-1626.5 MHz terminal uplink, spread and encrypted; the S-band satellite downlink is not tuned), QO-100 (beacon lock, narrowband and wideband transponders, DATV hand-over to DVB-S/S2) and CubeSat telemetry (AX.25 1200 / 9600 baud, AX100 modes, CCSDS Reed-Solomon, SatNOGS-style KISS out).

• Tone decoders (CTCSS, DCS, DTMF, five-tone, SELCAL and two-tone pages; details).
• DMR, NXDN (NXDN48 / 96), P25 (Phase 1, conventional and trunked), dPMR / PMR446 (all 16 channels scanned), D-STAR (also in the amateur group) and M17: call logs with IDs and talkgroups, data and short messages. P25 encrypted calls are only reported as encrypted. Voice for the AMBE / IMBE modes needs mbelib you install yourself, see Good to know.
• TETRA (base station broadcast: network and colour code, location area, carriers, services, neighbours, whether encryption is used; scan over a span) and TETRAPOL (D_SYSTEM_INFO). Receive only, no voice, nothing decrypted.

• Cell towers (LTE MIB / SIB1 - SIB5; towers on a map with an OpenCelliD key), 5G NR cells (SS/PBCH, PCI, RSRP, MIB; SIB1 is not decoded yet), NB-IoT cells, GSM-R (921 - 925 MHz) and LBJ (Chinese railway train warnings at 821.2375 MHz: train number, speed, kilometre post, direction; see docs/lbj.md). Public broadcast information only.
• KLW end-of-train data (default 866.2375 MHz) and TDCS train dispatch data (default 457.550 MHz): 1200 bit/s FFSK with BCH and CRC checks, locomotive and tail IDs, pressure / commands or train number, speed, mileage and consist fields. LBJ also decodes the fixed-position extension, WGS-84 coordinates and time broadcasts. Details and test coverage.

• POCSAG (512 - 2400 bd) and FLEX (1600 - 6400), whole span at once.
• Multitone Paging: a detector only (the format is not public); no message text is shown. Pager messages are private; receive only.

• ISM sensors (433.92 / 868 / 915 MHz: weather stations, tyre pressure, meters, remotes, Z-Wave; S0 and S2 stay encrypted).
• Wireless M-Bus (EN 13757-4 utility meters: S1, T1, C1 and R2 on 868 MHz, N1 on 169 MHz; manufacturer, ID, device type and the readings of unencrypted frames, encrypted payloads stay encrypted; details).
• LoRaWAN (EU868 / US915, all spreading factors; clear headers only, payloads stay encrypted), Meshtastic and MeshCore (nodes, map, chat, telemetry, exports), LoRa (raw) (plain point-to-point LoRa; hex and text frames, no decryption).
• Bluetooth LE (advertising, never connects), Zigbee / 802.15.4 (needs a fast radio), Wi-Fi networks (2.4 GHz beacons, passive), Wi-Fi HaLow (802.11ah sub-GHz beacons), Z-Wave (G.9959 R1 / R2 / R3, encrypted payloads stay encrypted), Wi-SUN FAN (smart meter and street light mesh, all regional band plans), ANT+ sensors and DECT cordless phones (base stations from beacons; calls are encrypted, no voice; Europe 1880-1900 MHz, US/Canada DECT 6.0 1920-1930 MHz, Brazil, Latin America and the US 900 MHz ISM band, see docs/DECT.md).

• Radiosondes (400 - 406 MHz and LMS6-1680; tracks and landing prediction), NOAA Weather Radio SAME (EAS alert decoder).
• Time signals (DCF77, MSF, WWVB, JJY, WWV / WWVH, BPM, BPC and more; the computer clock can be synced), e-Czas (Poland's official time on 225 kHz long wave, from a long-wave capable radio or an upconverter; how to receive it) and Ionospheric sounders (ionograms from chirp sounders).
• Radio astronomy (21 cm hydrogen line and OH lines).
• GOES / Meteosat DCS (river gauges, weather stations, tide gauges and buoys reporting on the 401.7 - 402.1 MHz uplink: every carrier in the capture followed, 100 / 300 / 1200 bps with NRZ or Manchester coding, the BCH-checked platform address, the channel and the ASCII or pseudo-binary sensor data, CSV logging; details).
• Meteor scatter (pings of meteor trails reflecting a distant beacon such as GRAVES on 143.050 MHz: start, duration, peak level, Doppler, rates per hour and day with the showers marked, CSV logging; details).
• BLE tracker detector (Apple Find My accessories such as AirTag, Tile, Chipolo, Samsung SmartTag and Google Find My Device tags recognised from their advertisements; how long and how close each stays, and an alert through the watchlists for one that follows you; receive only, identifiers kept for a short window and never saved; details).

• FSQ and IFKP keyboard chat (33 tones).
• ARDOP and SITOR are listed under HF utility.
• WSJT family: FT8, FT4, FT2, WSPR (with an optional spot upload to wsprnet.org), JT65, JT9, Q65, FST4 / FST4W, MSK144, and JS8Call.
• MT63 (500 / 1000 / 2000 Hz, short and long interleave found automatically, 64 carriers with Walsh-Hadamard coded characters; details).
• ARDOP (the Winlink HF modem, monitor only: frame types, connect requests, IDs and pings with call signs, ACK / NAK quality and the text of 4FSK data frames; PSK and QAM frames are listed by header; details).
• Keyboard modes: PSK31 / 63, RTTY, CW (every signal in a 14 kHz window), Olivia, Contestia, MFSK, THOR, DominoEX, Hellschreiber. Picture modes: SSTV and HamDRM (EasyPal).

• Yaesu Fusion (C4FM, in development): callsigns, call mode and slow data; voice only through a codec plug-in you install.
• FreeDV (codec2 included), M17, D-STAR, APRS (AFSK) / AX.25 packet (whole span at once), LoRa APRS and MeshCom.

• HF waveform ID (identify only): estimates the baud rate, tone spacing and repetition of a military or government HF signal and ranks candidate waveforms; nothing is decoded.
• SITOR (FEC-A and FEC-B, in development) and ARDOP (Winlink HF frames, monitor only).
• HF ALE (2G calls, soundings, LQA), HF beacons (NCDXF / IARU propagation monitor), Datawell Buoy HF Link (wave-buoy telemetry, HXV format decoded; the 4FSK HVA format is not public) and CODAR / HF ocean radar (sweeps measured; propagation only, no positions).

• KAAL voice and beacon (10 kHz, receive and transmit), KAPP packets (100 kHz), KAFT file transfer (200 kHz to 1 MHz), KAVB video (1 to 2 MHz, spec and physical layer only), KAIT pictures in 10 s bursts and KAWS HF weak-signal chat. Measured in simulation only, never on the air.

• eLoran / Loran-C (100 kHz, in development), MPT1327, LTR and EDACS trunking control channels (in development; control data only, no voice), GOES DCS (data collection platforms, 401.7 - 402.1 MHz), Mode A/C (squawk and altitude), BLE tracker detector (AirTag, Tile and others; warns about one that follows you), AM stereo (C-QUAM), DGPS beacon and DME (in development).

• General receiver: NFM, WFM, AM, USB, LSB and CW anywhere the radio tunes, up to 4 channels from one capture, filters, squelch, recording to WAV (one file per transmission), FLAC and MP3.
• Spectrum analyser (1 - 6000 MHz sweeps; CSV export), Spectrum explorer (markers, measurements, bookmarks) and Signal workbench (analyse and demodulate a recording from a file, no radio needed).

Takes a recording apart without a radio, in the spirit of Inspectrum, SigDigger and Universal Radio Hacker.

1. Open a file (or drop it on the window). Audio files (WAV, MP3, FLAC, OGG, M4A) are treated as a real signal; a stereo file can be read as I/Q ("stereo is I/Q"). IQ files (.cs8 .cu8 .cf32 .ci16 .cu16) are read from disk as they are, so 100 MB and larger files open at once; they need the sample rate (a name like capture_2Msps.cs8 gives it). SigMF and IQ WAV files describe themselves. "Demo signal" makes a recording to try the tools on.
2. Look at the waveform and a large spectrogram. Wheel zooms time, Ctrl + wheel zooms frequency, right-drag pans, a click puts the cursor, a drag selects a time range and a band.
3. Measure: peaks, bandwidth (-3 dB, -20 dB, 99 %), carrier offset, SNR, symbol rate, modulation guess, power over time and bursts.
4. Demodulate: AM, FM, SSB, CW, ASK/OOK, 2-FSK / 4-FSK, AFSK, BPSK / QPSK / 8PSK, MSK / GMSK, with eye diagram, constellation and the recovered bits; Listen plays the audio.
5. Read the bits: invert, differential, NRZI, Manchester, byte reversal and alignment, hex / binary / ASCII, pattern search.
6. Export the selection as .wav, .cf32 or SigMF, the audio as .wav, the bits as .txt or .bin, the peaks as CSV.

wbtool (built with the tests) does the same on the command line: wbtool gen out.cf32 --kind qpsk --rate 250000 --baud 12500, wbtool analyze file --rate R, wbtool demod file --kind fsk2 --baud 9600 --band f0 f1.

Cell towers broadcast who they are but not where they stand. The Cells mode can look each decoded LTE cell up in OpenCelliD, an open, crowd-sourced tower database, and put it on the map:

1. Sign up (free) at https://opencellid.org and confirm your e-mail.
2. Log in and copy your API token from your account page.
3. In OnAir, open Cells → Map and paste it into OpenCelliD API key.

Lookups need internet and run at most once a second; only the cell identity (MCC, MNC, TAC, cell id) and your key are sent. Answers are cached on disk, so the map also works offline afterwards. A CSV downloaded from OpenCelliD works too (Load tower file). Tower data © OpenCelliD, CC BY-SA 4.0.

In the Iridium screen, Save for iridium-toolkit writes every demodulated burst as the RAW: lines of gr-iridium's iridium-extractor (by default output.bits in the recordings folder). From the command line: onair-cli --standard iridium --iridium-bits output.bits --secs 600. Then run muccc/iridium-toolkit:

iridium-parser.py output.bits > output.parsed

OnAir writes the bits only and never decodes voice (the AMBE codec is patented).

• T2-Lite (1/3 and 2/5 code rates, short FEC frames, 1.7 MHz channels) is decoded like any other DVB-T2 signal and checked on simulated signals; not tried on a real T2-Lite broadcast. A T2-Lite signal sharing a channel with a normal T2 one (in its future-extension frames) is not decoded.
• ATSC 3.0 is built from the specifications and works on simulated signals (synchronisation, error correction, link layer, ROUTE, HEVC with AAC); details may need fixing against a real broadcast. If you have a recording, please tell us on Discord.
• ISDB-T follows ARIB STD-B31 (all modes and guard intervals, up to three layers, TMCC, interleavers, Viterbi, Reed-Solomon) and decodes simulated signals exactly; not checked against a real broadcast. Channel names in the Japanese character set are not converted yet. To try it without a signal, isdbtgen --out sample.cs8 (built with the tools) makes a recording: open it as an IQ recording, 10 Msps, CS8, with ISDB-T selected.

• Sound card (the radio list: Sound card: …, including Default input): WebSDR, a virtual audio cable, or a handheld radio's speaker/line into the computer, for HF decoders (FT8, RTTY, CW, SSB) at the capture rate (usually 48 kHz). Four mappings: I/Q stereo, USB audio, LSB audio, mono real. The frequency field is the dial the decoder should assume; nothing is tuned. Unverified against a real WebSDR.
• Hamlib rig (source list: Add Hamlib rig…): a CAT-controlled radio such as an Icom IC-R8600 or IC-7300. Search the model by name or type its Hamlib number, pick the serial port OnAir finds (or host:port for a rigctld on the network), and choose a sound card for the audio or IQ. The frequency and mode follow in both directions. If the rig is unplugged or rigctld stops, OnAir says so and reconnects by itself. Receiving never keys the transmitter. Rotators are driven the same way.

• rtl_tcp carries 8-bit samples at up to about 2.5 Msps (more with the Auto server radio setting). SoapyRemote needs SoapySDR with the SoapyRemote module on both computers; if it picks the wrong radio or lists none, type the radio's driver (for example mirisdr) in "Remote driver". Do not confuse SpyServer's port 5555 with SoapyRemote's 55132.
• SpyServer is spoken from the public protocol description (2.0.1700): the full IQ stream (8-bit or 16-bit), sample rates from the server's decimation stages, frequency and gain control. Tell us if your server misbehaves.
• SDR++ server: OnAir sets the sample rate and the gain through the server's own source menu, so it works with whatever radio SDR++ drives there, and that source's other controls (gain mode, bias tee, AGC, antenna, the server's source) show up under Radio settings as "SDR++: ...". Those are changes to the server: SDR++ keeps them after OnAir disconnects. The server takes one client at a time. See DEVICES.md.
• Hermes Lite 2 receives up to 38.4 MHz at 48 to 384 ksps, so only HF and narrow modes fit; the N2ADR filter board's band filters follow the frequency. Red Pitaya runs Pavel Demin's HPSDR-compatible SDR transceiver app: 48 to 384 ksps, tuning up to 61.44 MHz, no gain control; tested only against a fake server, so sample rate accuracy, level scaling and mirroring (setting "Swap I and Q") are unverified.
• Several receivers (slices) of one Hermes Lite 2 or Red Pitaya: the radio list has one entry per receiver ("Hermes Lite 2 RX1 ...", "Hermes Lite 2 RX2 ..."; as many as the HL2's gateware says it has, three on a Red Pitaya), so each OnAir receiver tab can take its own, on its own frequency and mode. They share the radio's one stream, so they share its sample rate (the first one opened sets it), its gain and the filter board (the filter for the highest frequency in use). Opening or closing a receiver above the others restarts the stream for a fraction of a second. One program at a time can stream from the radio.
• When a network radio drops, OnAir reconnects by itself, and the status bar shows the samples lost. Nothing is opened to the network by default.
• Share my radio (View > Data feeds): OnAir becomes an rtl_tcp server for the radio it is running, so a second OnAir, SDR#, SDR++ or any rtl_tcp client can listen to the same signal (8-bit IQ at the radio's rate). Only this computer can connect until you tick Let other computers connect (LAN); clients can change the frequency and gain only when Allow remote control is ticked.
• Data feeds (View > Data feeds), all off until you turn one on: ADS-B SBS-1 and Beast (TCP), AIS NMEA (UDP), APRS-IS and KISS (TCP), ACARS and VDL2 JSON (UDP), GNSS NMEA RMC/GGA (UDP), Bluetooth LE JSON (UDP), LoRa / LoRaWAN / Meshtastic JSON (UDP). The TCP servers stay on this computer unless Let other computers connect (LAN) is ticked.

Scripts and other programs can control OnAir over HTTP and WebSocket. It is off by default: turn it on in View > Data feeds > Remote control API. It listens on 127.0.0.1 (port 8765) only; Let other computers connect opens it to the network and then needs a token (at least 16 characters, made for you; the app shows it and can copy it). Programs on the same computer need no token; every other address must send Authorization: Bearer <token> (or ?token=<token>, which ends up in logs: prefer the header). A web page cannot drive it: no CORS headers are sent, a POST must be Content-Type: application/json, and a request with a foreign Origin is refused.

Replies are JSON; a refused or failed request gives {"error":"..."} with a 4xx / 5xx status. Changes are made on the program's main thread within a frame.

curl -s http://127.0.0.1:8765/api/v1/telemetry
curl -s -X POST -H 'Content-Type: application/json' -d '{"mhz": 474}' http://127.0.0.1:8765/api/v1/frequency
curl -s -H "Authorization: Bearer $TOKEN" http://192.168.1.20:8765/api/v1/status

The server is core/src/net/remote_api.cpp; tests/net/test_remote_api.cpp runs a socket client against it, and against the real engine on a generated DVB-T2 signal.

• The radio is not listed. Windows: install the WinUSB driver once with Zadig, see Install. SDRplay RSPs need the SDRplay API installed first. Notes per radio are in DEVICES.md.
• The picture breaks up. Use Auto-tune or Optimise gain, and move the antenna while you watch the quality score, see Quick start. A weak or echo-filled signal is the usual cause.
• macOS will not open the app. It is not signed with a developer certificate yet: right-click it and choose Open on the first launch.
• No voice in DMR, P25, NXDN, dPMR or D-STAR. Install mbelib yourself, see Good to know and docs/CODECS.md.
• No radio at all. Open an IQ recording through the IQ recording file source, or press Tour.
• Still stuck. Press Save bug report (one zip with diagnostics, log and crash report) and send it on Discord or in GitHub Issues.

You need a C++20 compiler, CMake 3.20 or newer, Rust (cargo and rustc 1.70 or newer; required since 0.2.7, the protocol layers are Rust), pkg-config, FFmpeg, libhackrf, and GLFW for the app. SoapySDR is optional. For a cross build install the Rust target first (rustup target add <triple>); CMake names it when it is missing.

# macOS
brew install cmake rust pkg-config hackrf soapysdr libusb ffmpeg glfw
# Debian / Ubuntu
sudo apt install build-essential cmake cargo pkg-config libhackrf-dev libsoapysdr-dev libusb-1.0-0-dev \
  libglfw3-dev libavformat-dev libavcodec-dev libavutil-dev libswscale-dev libswresample-dev

cmake -S . -B build -DCMAKE_BUILD_TYPE=Release -DDECT2_BUILD_APP=ON
cmake --build build -j "$(getconf _NPROCESSORS_ONLN)"   # one job per core (a bare -j starts hundreds and can run out of memory)
ctest --test-dir build          # run the test suite

The app is build/dect2. Without a radio, open an IQ recording through the IQ recording file source, or run the command-line receiver on it:

build/dect2cli --file capture.cs8 --rate 10 --format cs8 --play <service id>

SigMF (.sigmf-meta / .sigmf-data) and two-channel WAV IQ files open the same way and describe their own sample rate, centre frequency, format and start offset; --rate, --format and --freq still override.

Tuning for your own CPU. Release programs are built for the baseline 64-bit PC (the hot loops pick AVX2 at run time). A binary you build yourself can use everything your CPU offers:

• -DDECT2_WIN_AVX2=ON (x86-64 on Windows and Intel Macs): the whole receiver for AVX2 + FMA; refuses to start on older CPUs.
• -DCMAKE_CXX_FLAGS="-march=native" (GCC or Clang): runs only on CPUs like yours.
• -DDECT2_PORTABLE=ON (macOS, plain C++ instead of Accelerate) and -DDECT2_NO_SIMD=ON go the other way, for unusual CPUs, and are slower.

Configure with a clean build folder after changing these. Other options in CMakeLists.txt: DECT2_UI_BACKEND (Metal or OpenGL3), DECT2_WITH_SOAPY, DECT2_WITH_RFNM (when librfnm is installed: DEVICES.md).

Windows (10 or 11, 64-bit), step by step:

1. Install MSYS2 with its default settings, and open MSYS2 UCRT64 (not MINGW64: it has no HackRF and SoapySDR packages).
2. Update MSYS2 with pacman -Syu (if the window closes, open UCRT64 again and run it a second time).
3. Install the tools and libraries:

   pacman -S --needed git zip mingw-w64-ucrt-x86_64-{gcc,cmake,ninja,pkgconf,ffmpeg,glfw,hackrf,soapysdr,libiconv}

   Optional: mingw-w64-ucrt-x86_64-nsis for the setup .exe. (codec2 is built from third_party/codec2 with OnAir and goes into the zip.)
4. Get the source and build:

   git clone https://github.com/adswill/OnAir.git
   cd OnAir
   cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release -DDECT2_BUILD_APP=ON
   cmake --build build

5. Run it from the same shell, so Windows finds the DLLs: ./build/dect2.exe. Tests: ctest --test-dir build.
6. For a folder that runs on any PC, with all DLLs next to OnAir.exe, run tools/package/make_windows.sh: a portable zip in build-windows/, plus a setup .exe if NSIS is installed. From macOS or Linux, tools/package/make_windows_cross.sh builds the same package with a cross-compiler.

To use a HackRF, RTL-SDR or Airspy, install the WinUSB driver once with Zadig.

macOS app and disk image: run tools/package/make_dmg.sh. It targets macOS 15.0 by default (override with MACOSX_DEPLOYMENT_TARGET); every bundled library must support that version, and packaging stops if any executable or library requires a newer macOS.

Project layout