FreeDV is a suite of digital voice modes for HF radio. FreeDV technology is being developed by an international team of radio amateurs working together on Machine Learning, DSP, coding, design, user interface and testing.
For those wanting a video overview of what it is, you can watch https://www.youtube.com/watch?v=vv2ceThcJRg.
What it is
- FreeDV is a suite of digital voice modes for HF radio. Their flagship mode is the Radio Autoencoder (RADE). You can run RADE using a free GUI application for Windows, Linux, and OSX that allows any SSB radio to be used for high quality digital voice.
- RADE is a new mode with state-of-the-art performance, it is a contraction of Radio AutoencoDEr so-named because the modulation encoding uses a Machine-Learning method where the modulation and demodulation are achieved by training on a large number of speech samples over a modelled radio channel with typical propagation disturbances in phase and amplitude as found in HF radio. The speech is synthesised using the FARGAN neural vocoder (Frame-wise Auto-Regressive GAN) where GAN is a Generative Adversarial Network).
- Digital voice over HF SSB – so not a text mode like JS8Call or VarAC. It’s humans speaking to humans.
- With noise floors getting worse on HF, phone is getting more difficult to achieve, and this digital mode will help cut through the noise, much like DMR already does.
- Unlike DMR this uses 100% open source software including the CODEC.
- Think of it as Echolink, without the Internet in between, and it is not going via any Echolink servers.
How wide is it
This I found very interesting from the article in the December 2025 Practical Wireless magazine by Eric Edwards GW8LJJ, as it bodes well for cleaner signals with less interference. The following is an excerpt from the article:
The speech signal has an audio bandwidth of 8kHz, but the RADE V2.0.1 (‘Radio Autoencoder’) (latest FreeDV version at the time) signal requires just 1500Hz of RF bandwidth. The audio is centred at 1.5kHz with a low frequency limit of 700Hz and a high frequency limit of 2.2kHz. Even with this narrow setting the audio is equal to amateur radio FM quality.
This is much narrower than a standard SSB transmission. Another interesting fact is that there are no opposite sideband spurious signals because the lowest audio frequency is 700Hz from the CIO (Carrier Insertion Oscillator). If the CIO was not fully or greatly attenuated as would be the requirement for normal SSB practice, it would make no difference at all to the FreeDV reception as it is out of range of the audio signal. Tests have been carried out using both sidebands at the same time (duplex) with two stations, one on lower sideband and the other on upper sideband on the same frequency. There was perfect reception of both stations with no interference or overlap of the sideband modulations.
How to install it
Install the FreeDV software from the download section on https://freedv.org/. They also have some additional info on getting started at https://freedv.org/getting-started/.
How to set it up
First: sound card configuration
If you have a radio with a built-in sound card, ensure it is connected with the USB cable. On Windows, you should see the UART Bridge COM ports. This step is identical for JS8Call, VarAC, and similar digital modes.
Start up the FreeDV app, and go to Audio Config under the Tools menu (you could also try the Easy Setup option first): They split it into 4 different connections:
* Receive Radio to Computer
* Receive Computer to Speaker or Headphones
* Transmit from Microphone to Computer,
* Transmit from Computer to Radio
Everything can use the default sample rate of 48000. Make sure the computer to/from radio is set to USB Audio CODEC, and the others are set to whatever you are using for the computer microphone and speakers. Use the Record 2 Seconds and Play 2 Seconds buttons to test connections.
Second: the CAT/PTT control
Then go to CAT and PTT Config under the Tools menu: Note: Hamlib is built into FreeDV so you do not need to install or start it separately. I enabled Hamlib and set the Rig Model to my Icom IC-7100, using COM3 in my case. The radio address needs to be set and for my iCom that is 88 (this may be for iCOM radios only). This address is a hex address so for the IC-7100 with Hamlib model 3070 the CI-V address is 0x88 or 88 by default, the IC-7300 with model 3073 is CI-V address 0x94 or 94, and the IC-705 with Hamlib model 3081 is CI-V Address 0xA4 or A4. My serial rate for my iCom is 19200 and PTT uses CAT. There is a Test PTT button to test the setup.
Third: the Reporting configuration
Under Tools on the menu, select Options and the Reporting tab, to set your callsign and grid square locator. Ensure the reporting to PSK Reporter and FreeDV Reporter are enabled.
How to use FreeDV
Once that is working, you are ready to use FreeDV:
- Start with Mode RADEV1 for weakest signals (most noise).
- Under Report Freq at bottom right, you can switch bands.
- Press SPACE (for PTT) to toggle transmit on/off.
USB or LSB
On bands below 10 MHz, LSB is used for FreeDV. On 10MHz and above, USB is used. After much debate, the FreeDV community has adopted the same conventions as SSB, based on the reasoning that FreeDV is a voice mode.
At the bottom left will be an indicator showing LSB-D or USB-D (when Hamlib is controlling the radio). If that shows red, then your radio is using an unexpected mode (e.g. LSB on 20 meters).
Transceiver Filters
For most FreeDV use, your radio’s receive and transmit filters should be set to the widest possible (typically around 3 kHz). This allows easy switching between analog mode as well as the various FreeDV modes. There has not been any testing done to definitively confirm whether narrower bandwidths help with reception of FreeDV. Additionally, FreeDV already performs its own transmit filtering, so using additional narrow filtering on the radio will likely have little benefit (again, untested).
For reference, the channel widths of the currently supported modes are below:
| Mode | Width (kHz) |
|---|---|
| 1600 | 1.125 |
| 700D | 1.000 |
| 700E | 1.500 |
| RADEV1 | 1.500 |
Audio Processing
Sound card levels are generally adjusted in the computer’s Control Panel or Settings, or in some cases via controls on your rig interface hardware or menus on your radio. In-app adjustments can also be done by using the ‘TX Level’ slider at the bottom of the main screen; anything below 0 dB attenuates the transmit signal.
When FreeDV is running, you can observe the sound card signals in the main window tabs (From Radio, From Mic, To Speaker).
- On receive, FreeDV is not very sensitive to the From Radio level, adjust so it is mid-range and not clipping. FreeDV uses phase shift keying (PSK) so is not sensitive to amplitude.
- The transmit level from your computer to your radio is important. On transmit, adjust your level so that the ALC is just being nudged. More is not better with the FreeDV transmit signal. Overdriving your transmitter will lead to a distorted transit signal, and a poor SNR at the receiver. This is a very common problem.
- FreeDV 700D and 700E can drive your transmitter at an average power of 40% of its peak power rating. For example, 40W RMS for a 100W PEP radio. Make sure your transmitter can handle continuous power output at these levels, and reduce the power if necessary.
- Adjust the microphone audio, so the peaks are not clipping, and the average is about half the maximum.
Who is online for a QSO?
You can use any web browser and view active stations at https://qso.freedv.org/.