M17 Digital Voice: The Open-Source Amateur Radio Mode

Most amateur digital voice modes were created by manufacturers or standards bodies and built into commercial radios. M17 digital voice takes a different route. It is a digital voice and data protocol developed within the amateur community, with a specification that anyone can read and implement, and it is designed for the VHF band and above. That openness has produced firmware for existing radios, new hardware and software hotspots built by volunteers. This guide explains how M17 works, what you need to use it and how it compares with the better-known DMR, D-STAR and System Fusion systems.

What M17 Is

According to the project’s description on Wikipedia, M17 was developed by Wojciech Kaczmarski, SP5WWP, with contributions from the wider amateur community. It is primarily designed for voice on the VHF bands and higher, and the project has received grants from the Amateur Radio Digital Communications organisation. The specification and the reference code are published by the M17 Project, which also runs a news feed on its home page.

The idea is to give amateurs a digital mode that they can understand fully, extend and build equipment for without paying for licences or relying on a manufacturer. The protocol is therefore documented in public, and the source code of many tools is open. This makes it attractive for experimenters, for clubs that like to build their own repeaters and for anyone who wants to know exactly what happens inside the radio.

How the Signal Works

The technical description is compact. M17 uses frequency-division multiple access, which means that each stream occupies its own frequency channel, as in analogue FM and in most amateur repeater systems. The modulation is four-level frequency-shift keying at 4,800 symbols per second. The channel bandwidth is around 9 kHz, which fits within a 12.5 kHz channel spacing. The gross data rate is 9,600 bits per second, of which 3,200 bits per second carry the actual voice or data payload, and the rest is error correction and framing.

A transmission, called a stream, is divided into 40-millisecond frames, each starting with a 16-bit synchronisation word. Six frames form a superframe, which carries the link information. The protocol allows a low-speed data channel to run alongside the voice, for example to carry position data or a short message. The link information includes the callsign of the sender and the destination, so any receiver can identify who is transmitting without a central database.

Voice Coding with Codec 2

M17 uses Codec 2, an open-licence low-bit-rate speech codec developed by David Rowe, VK5DGR, and others. The open licence is one of the reasons for its choice: nobody has to pay a royalty or sign an agreement to build a transmitter or a receiver. The protocol supports two voice rates, 3,200 bits per second for full rate and 1,600 bits per second for half rate. In the half-rate mode, the remaining capacity can carry a second data channel next to the voice.

The sound of low-bit-rate voice codecs is often described as slightly synthetic, and M17 is no exception. In practice, the audio is clear at good signal levels and degrades as the signal weakens. Developers on the project keep improving the receiving side, and recent news from the project describes experiments with a neural decoder that produces a wider-band output from the same transmitted bits. Such work shows an advantage of an open system: a new receiver can improve quality without any change to the transmitter, and every existing radio on the air benefits from the better decoder as long as it sends the standard stream.

Callsign Addressing Without a Central Registry

A distinctive design choice is the way stations are identified. M17 encodes the callsign directly in a 48-bit field that can hold up to nine alphanumeric characters. That means a radio does not need a user identity number from a central database, which is a requirement in some other digital voice systems. Your callsign is your address, and you can program it into the radio and a hotspot yourself. The system is simple to administer and respects the traditional way amateurs identify themselves on the air.

The protocol also supports a destination field, so a transmission can address a specific station, a group or a reflector. This is the way networked operation is organised, and it is similar to how amateurs use talk groups in other systems, although the details differ. The reflectors and gateways are run by volunteers, and the operators keep their own lists and rules, which you should read before you join.

Stream Mode, Packet Mode and Extras

M17 has two modes of operation. Stream mode is for continuous transmissions such as voice. Packet mode is for discrete data such as text messages, position reports, AX.25 frames and other binary data. Stream mode can deliver one 3,200 bits-per-second channel for voice or data, or two 1,600 bits-per-second channels, for example voice together with data. Packet mode supports text messaging, APRS, AX.25, and gateways to other protocols. The mode has also been demonstrated through geostationary satellite transponders, including the amateur transponder described in our guide to the QO-100 satellite.

The protocol also defines optional security features. These include authentication by digital signature, which proves that a transmission really came from the claimed station, and scrambling and encryption options. Be careful here. Amateur regulations in most countries do not allow transmissions that are encrypted to hide their meaning, so the encryption features are not for use on the amateur bands in those countries. Digital signatures, which do not hide the content, are a different matter. Always check your own licence conditions before you use any feature that goes beyond plain, identifiable communication.

Hardware for M17

You have several routes into the mode, and the market is changing quickly, so check current information from the project before you buy. Wikipedia notes that a few handheld transceivers can be flashed with open-source firmware to support M17 after a small hardware modification, and that in July 2024 a commercial handheld with native M17 support was released. Project hardware designs and open firmware such as OpenRTX are also available for people who want to experiment. A hotspot, built from a small computer and a low-power radio module, links your own radio to the network of reflectors in the same way that hotspots work for other digital voice systems; see our guide to an MMDVM hotspot for a related setup.

Software options also exist. Phone and computer clients let you listen and talk through a reflector without any radio at all, and the project publishes experimental versions that test new decoder improvements. These clients are a good way to try the mode and to hear what it sounds like before you spend money on hardware. Treat experimental software with the care that you give any early release: expect bugs, and check the sources of downloads.

Networks and Bridging

M17 networks are made up of reflectors, which act as shared rooms, and gateways that connect them to repeaters and hotspots. Traffic can also be bridged to other systems. According to Wikipedia, links exist between M17 and several other modes and networks, including DMR, P25, System Fusion, D-STAR, NXDN, AllStarLink, EchoLink and IRLP. Bridges are run by volunteers and can change, so check what is currently available. If you are interested in linking systems in general, our articles on AllStarLink and EchoLink show how other internet-linked systems are organised.

When you use a bridge, remember that the audio passes through several conversions, and each one can reduce quality. Choose a native reflector for the best result, and use bridges to talk to friends on other systems. Follow the rules of the network, and keep your transmissions short and clearly identified.

How M17 Compares with Other Digital Voice Modes

For a detailed look at the established systems, see our comparison of DMR, D-STAR and System Fusion. The main practical difference is the size of the community. DMR and the other systems have more radios and repeaters, while M17 offers openness and a fast pace of development.

Should You Try M17?

If your main goal is to talk to a lot of people in your region today, the established digital systems or plain FM repeaters will probably serve you better, because that is where the activity is. If you like to experiment, build and learn, or if you value open standards, M17 is a very rewarding project. You can start with a client program, listen to the reflectors, and then decide whether to add a hotspot or a radio. Ask your local club whether anyone is already active; the mode grows through small groups of enthusiasts who agree to try it together.

Keep expectations realistic. The network is smaller, quality depends on your receiver and equipment support is changing fast. On the other hand, you are part of a community that shapes the mode, and your reports and suggestions can influence its direction. Many amateurs find that this involvement is the best part of the project.

Getting Started: A Simple Path

First, read the project’s pages and the documentation. Second, check your licence rules for digital voice and for networked operation in your country. Third, try a client on a computer or phone and listen on a busy reflector. Fourth, if you want to transmit over the air, choose a hotspot or a radio that is currently supported, and follow the setup guide. Fifth, set your callsign carefully in the configuration, because errors in the identification field create confusion on the network. Finally, say hello, keep your calls short and join a regular net if there is one.

Etiquette on M17 Reflectors

Reflectors are shared spaces, so the usual rules of good operating apply. Listen before you transmit, because someone may already be talking. Give your callsign at the start and end of a transmission as your licence requires, keep over-length conversations to a net or a quiet room and avoid playing music or recordings. Many reflectors have module letters that work like separate rooms, so pick one that suits your topic. If you are testing, say so, and leave a pause so that others can break in. These habits make a small network pleasant and encourage newcomers to stay.

The Bottom Line

M17 digital voice is an open, amateur-built protocol that uses four-level FSK, the open Codec 2 voice codec and callsign addressing built into the protocol. It is smaller than DMR, D-STAR and System Fusion, but it is growing, it has hardware and software options, and it appeals to experimenters who value transparency. Check the current state of equipment and networks on the project’s website, follow your licence rules, avoid encryption on the amateur bands where it is not allowed and start with a software client to see whether the mode suits you.

Frequently Asked Questions

What does M17 stand for?

M17 is the name of the project and the protocol. The name itself is just a label chosen by the developers, and what matters is that the specification is open and developed in the amateur community.

Which voice codec does M17 use?

M17 uses Codec 2, a low-bit-rate speech codec with an open licence. The protocol supports a full-rate mode of 3,200 bits per second and a half-rate mode of 1,600 bits per second, where the spare capacity can carry data.

Do I need a special radio for M17?

You need a radio that supports the mode, which can be a commercial handheld with native support, a radio running suitable open firmware or a hotspot linked to a standard FM radio. You can also listen and talk through a software client without any radio. Check the project’s site for current options.

Can I use encryption on M17?

The protocol defines scrambling and encryption options, but amateur regulations in most countries prohibit encryption intended to hide the meaning of a message. Check your own rules; digital signatures that authenticate a transmission without hiding it are a different feature.

How does M17 compare with DMR?

DMR is a mature open standard with a huge number of radios and repeaters, and it identifies users by an ID from a database. M17 is newer and smaller, uses one stream per channel and puts the callsign directly into the protocol. DMR has more activity today, while M17 offers openness and rapid development.

Can I link M17 to other digital voice systems?

Volunteer-run bridges exist to several other modes and networks, but they change over time and add audio conversions that reduce quality. Check which bridges are active and follow the network rules. For the best sound, use a native M17 reflector.