The only thing missing is an audio transmission between your radio and Jonathan's radio to get an idea of what the sound quality is between your stations on FM with the analog signal
No doubt room for improvement. I do believe that P25 is far better, of course using Motorola XTL5000 radio's, on VHF/UHF. Not sure if their protocol is available/open source. I must say, a few years ago, I came across a fellow in Australia that was working on a HF digital voice project, developing the code, a few years ago, seemed promising, unfortunately can't recall who he was, nor could I find the documentation on my computer. Could be possibly, in QEX?? Good work as usual Mat, you are definitely an asset to the Amateur Radio Community! I've been a ham since the early 70's, (Rf Engineer since '80's, radio has been my life since the late 60's!) nice to see that someone is taking the time to inform/educate the radio community, and that you have a following. A lost tradition over the decades! You dedicate plenty of time in doing so, my respects to you, and by the way, as usual, thumbs up! HI-HI! 73's Dale ve3aam
Interesting comparison. You can understand what's being said and can identify the speaker. All sound robotic, like over-compressed MP3. You pay your money and you get your bit rate. 🙂 I work with VoIP systems and we use G.729 as our codec. Now that the patents have expired there are no licensing issues.
We used to run G.729 when bandwidth was a real issue. Now it's boring G.711 or G.722 (or Opus for internal calls if I'm feeling it). You could 100% fit G729 in 5khz bandwidth... But also lose all interoperability with normal radios.
Well done Mat and Jonathan! Current state of the DV art for hams is not quite viable. As newer technology comes available (obsolete for cell phones, but good for us), I think we will see dramatic improvements. As I said on Mat's channel, compare the first digital cameras with what we have now. Remember all the guys who said film will always be superior? 73, Len, KD0RC
I do like to hear it on using it at SSB, both as a strong signal, and with a low SNR, to hear how godt it will perform on a god and low SNR, with the limited band with ??
There are already too few digital mode operators, and now when you want to operate with freedv you have to make a choice between incompatible modes 🤷♂️
I have done a fair bit of playing with FreeDV in the past, and for whatever reason, my results appear to be a lot better than yours here. And interestingly, I was doing this over SSB on HF. Maybe a re-test is needed on this one Matt? I have found it to be a really fun and impressive mode, but unfortunately this video paints the FreeDV mode in a fairly bad light, possibly unfairly.
I totally agree. I would give it a try on SSB, too. I suspect that setting the input for “no ALC level” on TX (as for FT8 and other digital modes) would help; in general, a linear, unfiltered/uncompressed signal is needed, both on TX and RX. My extremely limited experience seems to show that better quality is definitely achievable. Unfortunately, I miss FreeDV activity here in Europe to test and learn together, on the bands. Thanks for your videos Matt, they’re always fun and instructive! All the best and 73, Pat, HB9FAB
Yep, called ezDV. It'll support 700D, E and 1600 (similar to the existing SM1000 device that Tech Minds alluded to at the beginning of his video). No firm pricing yet but yeah, assuming nothing goes wrong, it'll hopefully be out in a few months.
Ever since Dstar came out I’ve wondered why anybody would want their voice to sound that bad. These modes just demonstrated are no better and I sure can’t see any reason for their use.
The problem with most digital modes is the instability of audio frequencies during vowel sounds. This makes it extremely confusing and cognitively demanding to listen to. The 2020 was maybe a bit better in that respect.
Most are optimized for male voices as well and the higher frequency harmonics more typical in female voices are crushed by ambe codecs etc. very cool to compare these!
horrible decoded audio even in 2020 mode. I did a similar exercise a few years ago on 10M SSB with similar results. Basically, the analogue audio is much better than the decoded digital even with noise present. What I noticed too is just how strong the signal has to be to get good copy. I'm no expert but the 2020 modes sounded like there had been some data loss and that's at S5 on FM when I'd expect to get perfectly readable analogue audio. I've been following the progress of FreeDV and Codec 2 for years and I really want this stuff to work properly but, to my tired old ears, it's still not there yet
Problem is, those codecs producing garbage audio at those low data-rates, EnCodec with Multi-Band Diffusion (MBD-EnCodec) sounds fascinating at 3Kbps, requires about 4kbit RAW datarate with FEC and so on, using DQPSK, or a smaller RF bandwidth D8PSK modem.
The only thing missing is an audio transmission between your radio and Jonathan's radio to get an idea of what the sound quality is between your stations on FM with the analog signal
There are audio loopback examples available within GNU Radio Companion for both FreeDV and Codec2.
Fascinated and definitely will be looking into this! Thanks!
No doubt room for improvement. I do believe that P25 is far better, of course using Motorola XTL5000 radio's, on VHF/UHF. Not sure if their protocol is available/open source.
I must say, a few years ago, I came across a fellow in Australia that was working on a HF digital voice project, developing the code, a few years ago, seemed promising, unfortunately can't recall who he was, nor could I find the documentation on my computer. Could be possibly, in QEX??
Good work as usual Mat, you are definitely an asset to the Amateur Radio Community! I've been a ham since the early 70's, (Rf Engineer since '80's, radio has been my life since the late 60's!) nice to see that someone is taking the time to inform/educate the radio community, and that you have a following. A lost tradition over the decades! You dedicate plenty of time in doing so, my respects to you, and by the way, as usual, thumbs up! HI-HI! 73's Dale ve3aam
can u test on hf band in SSB?
Love your videos. Keep up the good work. Are you working on a review of the the new tidradio td-h3?
Interesting comparison. You can understand what's being said and can identify the speaker. All sound robotic, like over-compressed MP3. You pay your money and you get your bit rate. 🙂
I work with VoIP systems and we use G.729 as our codec. Now that the patents have expired there are no licensing issues.
We used to run G.729 when bandwidth was a real issue. Now it's boring G.711 or G.722 (or Opus for internal calls if I'm feeling it).
You could 100% fit G729 in 5khz bandwidth... But also lose all interoperability with normal radios.
@Ardren how would you fit an 8kbps codec into 5khz? Would you use OFDM?
To me, this sounded more like a data loss/corruption demo than a test of encoding methods.
Appreciate the info and research.
Exactly what I was looking. Thanks. Now searching for setting up. Because on HF it is terrible. But I just started. 73 de TA2OM
Can I use it with rtl-sdr to pure decoding incoming signals?
Honestly I struggled to make out what he said on most of those but 2020.
For vhf/uhf I’ve had much clearer audio via FM or even YSF
yes 2020 is the only DV mode worth bothering with I guess. All other modes sound worse than analogue landline still in 2024
yes 2020 is the only DV mode worth bothering with I guess. All other modes sound worse than analogue landline still in 2024
Well done Mat and Jonathan! Current state of the DV art for hams is not quite viable. As newer technology comes available (obsolete for cell phones, but good for us), I think we will see dramatic improvements. As I said on Mat's channel, compare the first digital cameras with what we have now. Remember all the guys who said film will always be superior?
73,
Len, KD0RC
I wanted to use this on fm. The developer made for hf modes. But he said would work. I know one of the project people on freedv.
Be interesting to see that same test with a more stronger signal between both stations and different modes ssb and FM, 2020 did sound better.
I do like to hear it on using it at SSB, both as a strong signal, and with a low SNR, to hear how godt it will perform on a god and low SNR, with the limited band with ??
Both released your videos n nearly at the same time
There are already too few digital mode operators, and now when you want to operate with freedv you have to make a choice between incompatible modes 🤷♂️
2020C sounds best, but it's removed in the acutal versions, didn't had this garble with 2020C compared to 2020
To everyone commenting, don't like it. Move on, it's not for you.
Yes I can hear you Clem Fandango! :)
I have done a fair bit of playing with FreeDV in the past, and for whatever reason, my results appear to be a lot better than yours here. And interestingly, I was doing this over SSB on HF. Maybe a re-test is needed on this one Matt? I have found it to be a really fun and impressive mode, but unfortunately this video paints the FreeDV mode in a fairly bad light, possibly unfairly.
I totally agree. I would give it a try on SSB, too. I suspect that setting the input for “no ALC level” on TX (as for FT8 and other digital modes) would help; in general, a linear, unfiltered/uncompressed signal is needed, both on TX and RX. My extremely limited experience seems to show that better quality is definitely achievable.
Unfortunately, I miss FreeDV activity here in Europe to test and learn together, on the bands.
Thanks for your videos Matt, they’re always fun and instructive!
All the best and 73,
Pat, HB9FAB
Freedv is developing a new board it will be available in a few no months. For about $100. Use with out internet.
Yep, called ezDV. It'll support 700D, E and 1600 (similar to the existing SM1000 device that Tech Minds alluded to at the beginning of his video).
No firm pricing yet but yeah, assuming nothing goes wrong, it'll hopefully be out in a few months.
Odd that multiple U Tubers all post similar subject matter videos within the same day/s as each other 🧐
Ever since Dstar came out I’ve wondered why anybody would want their voice to sound that bad. These modes just demonstrated are no better and I sure can’t see any reason for their use.
Slightly choppy and garbled!
Thats a understatement.. 😅
Digital voice has loooong way to go to be honest. Hard to listen to for any length of time. Never used DStar or DMR.. they sound the same?
Apart from the 2020, they are all extremely tiring to listen to.
Even the 2020 is only just passable and in the same conditions FM would be easier on the ear
They all sounded equally poor to me. Needs lots of improvement to become something for every day use.
The problem with most digital modes is the instability of audio frequencies during vowel sounds. This makes it extremely confusing and cognitively demanding to listen to. The 2020 was maybe a bit better in that respect.
Most are optimized for male voices as well and the higher frequency harmonics more typical in female voices are crushed by ambe codecs etc. very cool to compare these!
None of these modes are acceptable for me.
horrible decoded audio even in 2020 mode. I did a similar exercise a few years ago on 10M SSB with similar results. Basically, the analogue audio is much better than the decoded digital even with noise present. What I noticed too is just how strong the signal has to be to get good copy. I'm no expert but the 2020 modes sounded like there had been some data loss and that's at S5 on FM when I'd expect to get perfectly readable analogue audio. I've been following the progress of FreeDV and Codec 2 for years and I really want this stuff to work properly but, to my tired old ears, it's still not there yet
Problem is, those codecs producing garbage audio at those low data-rates, EnCodec with Multi-Band Diffusion (MBD-EnCodec) sounds fascinating at 3Kbps, requires about 4kbit RAW datarate with FEC and so on, using DQPSK, or a smaller RF bandwidth D8PSK modem.
None.
Is anyone going to make a digital voice mode that doesn't sound like crap?
A solution searching for the problem.