Codec2 - Next-Generation Digital Voice for Two-Way Radio

The American Radio Relay League has awarded David Rowe the 2012 ARRL Technical Innovation Award for his work on Codec2.

Hams like you are again leading development in telecommunications, not following commercial developments. Codec2 offers very significant advantages over any voice system available for Amateur, Commercial and Emergency Services communications.

• Telephone-quality voice, comparable to industry-standard codecs used on landline, cell phones, and VoIP while using 1/8 of the bandwidth or less.
• Using the FDMDV-derived softmodem:
  • 1.1 KHz bandwidth, half the bandwidth of an SSB channel!
  • Works well on existing SSB radios, using the same set-up as other digital modes.
  • Can be carried on FM audio of existing VHF/UHF radios.
• Using a GMSK or C4FM modem for VHF/UHF
  • Potential for 2 KHz channels, given sufficient frequency precision in mobiles and handhelds.
  • 2.5-3 KHz channels are easily possible with today's designs.
  • Less than half the bandwidth of D-STAR, less than 1/4 the bandwidth of Yaesu DMR/TRBO systems.
  • 1/2 to less than 1/4 the bandwidth of APCO P25 systems, even Phase 2 and CQPSK.
  • At least double the power efficiency (and thus the range) of wider-bandwidth modes.
  • Can use non-linear amplification, for best battery life.
• Fully legal for international communications because it is not an unspecified digital code under FCC rule 97.309(b) or laws of other nations.
• Multiple available bit rates, currently between 1200 and 2400 bits per second, with increment in voice quality at higher rates.
• The codec is separate from modem, modulation and error-correction, choose the best modulation and error-correction for your channel.
• Open Source code available for download, for both the codec and the soft-modem.
• Avoids intellectual-property issues common to codecs:
  • No royalties, liberal grant of copyright rights to everyone.
  • Engineered to avoid existing patents in the field.
  • Exportable to all nations: "public domain" information under U.S. ITAR 120-125, found unrestricted in a letter of determination from the Australian government.
• Any Amateur Radio manufacturer can use Codec2, no charge, no agreements to sign.

Please join us! Subscribe to Our Announcement List so that we can notify you of new Codec2 software and events.

You can listen to the received signal, over a real 925 mile HF path. This is an alternating 1/0 bit pattern to test the modem, we'll put up a signal containing encoded audio ASAP. The image is a spectrogram (waterfall plot) of the same signal. Time is along the x axis, frequency along the y axis. The “hotter” the colour, the stronger the signal. Our FDM signal is the parallel red lines between 600 and 1700Hz. Above the modem signal is some analog SSB. You can hear this as the high frequency “Donald Duck” sound in the received signal. Now around 2.5 and 3.3 seconds there are strong bursts of SSB right on top of our signal, in the 0 to 1100 Hz range.

David Rowe VK5DGR is the codec and modem software developer.

These pages provide information on Codec2 and the FDMDV modem:

• Video: David Rowe explains Codec2 at Linux.conf.AU 2012
• Codec2
• Codec2 at 1400 bits-per-second.
• FDMDV
• README for FDMDV source code.
• Testing an FDMDV Modem
• Source Code.
• Public Mailing Lists at lists.codec2.org .
• Developers mailing list at Sourceforge.
• People who will give a speech about Codec2 at your hamfest or club.
• How to promote Codec2 at your hamfest or event.
• Chat room.

Why Have We Developed Codec2, and How Have We Been Able to Do It?

Several years ago, a number of hams became concerned by the introduction of “black box” codecs to Amateur Radio. For a century of ham radio, there has never been a barrier to understanding of any part of it, and a ham could build any part of it on his or her own – until digital audio and some digital modems were introduced. The way that current digital voice products for Radio Amateurs encode voice signals to digital bits is both trade-secret and patented. You're not allowed to understand how it works or duplicate it.

So, we decided to make our own codec, and make it Open Source, so that everyone could understand it, copy it, put it in software or hardware, build and improve it. The industry approach to codecs, though, is that they can only be understood by an elite few at companies that protect their secrets like gold in Fort Knox, and require pernicious legal agreements and a fortune in money just to look at them.

We were lucky to find David Rowe VK5DGR as our main algorithm developer. David has a Ph.D. in voice coding, and has shown that so-called “magic” algorithms are really just conventional math. David previously implemented OSLEC, the Open Source Line Echo Canceller, replacing the highly proprietary algorithms used to remove echo from digital voice telephony with a free implementation of excellent quality. David designed the IP04: Open Source PBX hardware that is sold commercially, and the Village Telco: an experimental WiFi mesh telephone network that provides free telephone service to people in Dili, East Timor.

In his work on Codec2, David has improved upon the current state of the art, delivering voice quality equivalent to industry standards, but at much lower bandwidth. All in Open Source software, free for use in commercial products or homebrew. Although developed by and for hams, Codec2 will revolutionize commercial and public service two-way radio, and offers the cellular telephone industry a chance to carry many more voice channels per MHz than they do today.

David has now re-implemented the FDMDV digital voice modem in Open Source software, with assistance from Mel Whitten K0PFX and many others, and is testing it on HF. Dave Witten, KD0EAG is leading development of a free multi-platform application using Codec2 and the FDMDV modem, to deliver digital voice to today's HF SSB operators.

When will Codec2 make it to VHF/UHF? It's possible to experiment today, using common FM radios and a computer. Deployment on smartphones and embedded processors must wait for the development of a version of Codec2 using fixed-point math, rather than the floating-point of the current implementation. There is no technical barrier to the use of fixed-point in Codec2's algorithms. Handheld radios, with their tight power budgets and non-linear amplifiers, probably call for a different sort of modem than we'd use on HF, for example GMSK. This technology is well-known.

Soon enough, it will be desirable to manufacture VHF/UHF radios that are purpose-built for the use of software codecs like Codec2, and that can profit most from its low bandwidth. The best hardware implementations to run Codec2 will offer user-programmable DSP, whether base, mobile, or handheld. Current digital voice radios all make use of DSP, they simply don't offer user programmability because they are intended for use with secret algorithms only. We will define standard versions of Codec2 for broad compatibility across manufacturers. But we believe that the technical limits of Codec2 have not been reached, and that we can drive its minimum bandwidth below 1000 bits per second, as well as to greater voice quality at higher bit rates. The next wave of digital voice will include rapid change as algorithms improve, and thus reprogrammable DSP will be essential.

Codec2 voice will become a major HF mode by the end of 2012. We expect to demonstrate purpose-built VHF/UHF implementations at Hamvention 2013.