FreeDV 700C

Over the past month the FreeDV 700C mode has been developed, integrated into the FreeDV GUI program version 1.2, and tested. Windows versions (64 and 32 bit) of this program can be downloaded from freedv.org. Thanks Richard Shaw for all your hard work on the release and installers.

FreeDV 700C uses the Codec 2 700C vocoder with the COHPSK modem. Some early results:

  • The US test team report 700C contacts over 2500km at SNRs down to -2dB, in conditions where SSB cannot be heard.
  • My own experience: the 700C speech quality is not quite as good as FreeDV 1600, but usable for conversation. That’s OK – it’s very early days for the 700C codec, and hey, it’s half the bit rate of 1600. I’m actually quite excited that 700C can be used conversationally at this early stage! I experienced a low SNR channel where FreeDV 700C didn’t work but SSB did, however 700C certainly works at much lower SNRs than 1600.
  • Some testers in Europe report 700C falling over at relatively high SNRs (e.g. 8dB). I also experienced this on a 1500km contact. Suspect this is a bug or corner case we can fix, especially in light of the US teams results.

Tony, K2MO, has put together this fine video demonstrating the various FreeDV modes over a simulated HF channel:

It’s early days for 700C, and there are mixed reports. However it’s looking promising. My next steps are to further explore the real world operation of FreeDV 700C, and work on improving the low SNR performance further.

6 thoughts on “FreeDV 700C”

  1. Hi David and all,

    I suppose the feature is already there? But how about automatically recording the digitized stuff from both ends of the link? Might that help understand and locate whatever corner cases or other problems there might be when an 8db link fails to work?

    John

    1. Yes already an option to record the received modem signal from the radio under the FreeDV GUI Tools menu. I’m also adding some more instrumentation, e.g. to plot BER per carrier and count resyncs.

  2. Hi David,

    I continue to be amazed how a much a simple to use plot tool helps me to understand what is really going on. I keep wondering what sort of easy to use software tool might be added to make this sort of analysis even easier?

    My typical use seems to be a dual trace O-scope looking at say data sent and data received or data measured one way verses another way. Easy to zoom in and out, make the vertical axis bigger and smaller, view live data or view recorded data etc.

    David, I continue to be amazed with your work :).

    warm regards,
    John

  3. I have been testing out the latest development release, I created a simple script to take mp3 files and create a flac file encoded with the output of a given codec2 mode. I figured I would take podcasts, and LibriVox audiobooks as test cases as they are optimally recorded and showcase distinct voices. What I found is that 700C is good in many cases but it has a noticeable regression on female voices / soft voices. The intelligibility drops in some cases to be unintelligible in 700C while intelligible in the 2400 mode. Some British accents don’t fair well either.

    I think that LibriVox is a great testbed for long form listening as it is public domain, has a variety of voices and has verifiable text if one wanted to do intelligibility tests. I am amazed at what has been accomplished so far and the quality possible at 700 bit/s. I look forward to tracking the amazing progress, best of wishes.

    1. Fine Business Joseph – nice work with your tests. It would be good to dig a little deeper into 700C and find out what part of the algorithm is struggling on your test samples. For example we could disable the various quantisation steps until the samples sound good again. Or print out objective measures for each sample, for example the average spectral distortion of the vector quantiser.

      I’m happy to help you (or anyone else) dig a little deeper into how these samples are interacting with 700C.

      – David

      1. I am happy to help test, and welcome others as well. I will see if I can get a short sample that crams a bunch of different voices in one clip so it is easier to spot regressions.

Comments are closed.