FreeDV is the fully open-source alternative to proprietary digital voice modes like D-STAR or DMR. Where those rely on the licence-encumbered AMBE codec and dedicated radios, FreeDV runs on ordinary SSB hardware and is open from the voice coding all the way to the modem. This article explains how FreeDV works, what the underlying Codec2 delivers, why the new neural mode RADE lifts voice quality to a new level, how to get started, and where there is activity in Austria.
What is FreeDV?
FreeDV (Free Digital Voice) is an open protocol for digital voice transmission on shortwave. It was developed by the Australian radio amateur and communications engineer David Rowe, VK5DGR. Its appeal lies in its openness: unlike D-STAR, DMR or C4FM, there is no proprietary vocoder chip you have to buy behind it — codec, modem and software are completely open-source and free of charge.
At its heart is the open-source voice codec Codec2, also by David Rowe. It squeezes speech into extremely low data rates (around 700 to 3200 bps) — little enough to transmit it robustly over a narrow SSB channel on shortwave. So you don't need a special digital radio: an ordinary SSB transceiver plus a PC and sound card is enough.
RADE: the big leap in voice quality
The most important innovation of recent years is RADE (Radio Autoencoder). RADE is a neural codec, developed by David Rowe together with Jean-Marc Valin (well known from Opus). It combines machine learning with classical signal processing and thereby delivers voice quality that is clearly superior to classic digital voice on shortwave and in places even approaches SSB — with an audio bandwidth of 8 kHz that occupies only about 1500 Hz of RF bandwidth. Its robustness is impressive: RADE still works at signal-to-noise ratios of around −2 dB, that is, where conventional methods have long given up.
RADE V1 arrived with FreeDV 2.0.0 in June 2025 and is today the flagship mode. Anyone who has avoided FreeDV so far because of its somewhat "robotic" voice quality should give it another listen at the latest now.
The FreeDV modes at a glance
- RADE: the neural mode (2025) — best voice quality, very robust down into weak signals. The recommended starting point.
- FreeDV 700D / 700E: 700 bps, extremely robust — for weak signals and poor conditions, when quality is secondary.
- FreeDV 1600: the older standard HF mode with usable quality, today used mainly for compatibility reasons.
- FreeDV 2020 / 2020B: earlier neural modes (LPCNet) for good quality — now largely superseded by RADE.
Note: the 2400A/B modes are intended for FM operation on VHF/UHF, not for classic SSB operation on shortwave.
Setting up FreeDV
Getting started is straightforward and inexpensive. You need three things: an SSB transceiver, a PC and an audio interface between the two (a SignaLink USB, DigiRig or RigBlaster are common; at a pinch the internal sound card will do). The FreeDV GUI software is available free of charge for Windows, Linux and macOS at freedv.org.
It is quickly set up: connect FreeDV to the sound card, optionally add the transceiver's CAT control, and handle transmit switching (PTT) via VOX or a serial line. Watch for good microphone levels (don't overdrive) — the rest, namely the actual coding, is done by the software. The getting-started guide walks you through the settings step by step.
Operating technique
Operation is in USB (upper sideband); the FreeDV software generates the digital signal from it. A CQ call sounds, for example, like this (example callsign):
"CQ CQ CQ FreeDV, this is OE0ABC, OE0ABC, over."
The classic FreeDV calling frequency is 14.236 kHz (20 m); alongside that, activity is common on 7.177 kHz (40 m) and 3.643 kHz (80 m). As always: listen briefly to check whether the frequency is clear, mind the band plan, and bear in mind that the exact calling frequencies can shift over time.
FreeDV in Austria
In Austria, FreeDV runs mainly on 40 m (7.177 kHz) and 20 m (14.236 kHz). Activity is still modest but growing — and the new RADE mode should give interest an additional boost. The appeal for OE is obvious: no licence fees (unlike the VARA modem), a fully open standard, and it works with any SSB rig that is already in the shack. The easiest way to find other stations is to listen on 14.236 kHz at the usual times or arrange a sked.
FreeDV compared
- FreeDV: open Codec2/RADE, HF-capable on any SSB rig, PC and sound card required, no fees.
- D-STAR / DMR / C4FM: proprietary AMBE codec, dedicated radio required, mainly on VHF/UHF and over networks/reflectors.
- VARA / Winlink: data modem (text/e-mail), not voice — and VARA HF is partly paid.
FreeDV is therefore the choice for anyone who wants digital voice on shortwave without being tied to a proprietary codec or a particular device.
Videos: FreeDV in practice
Three tutorials, from an overview through setup to on-air operation:
Further sources
- freedv.org — official project site, FreeDV GUI, modes and frequencies
- freedv.org/radio-autoencoder — background on the neural RADE mode
- freedv.org/getting-started — step-by-step introduction
- github.com/drowe67/radae — source code of the Radio Autoencoder (David Rowe)
73 – the oeradio.at team
Transparency Notice
This article was researched and written with the assistance of AI (Claude, Anthropic). The content was reviewed by the oeradio.at editorial team and prepared for the Austrian amateur radio community.
