Australian Ham Radio Discussion Forum ( AHRDF )

Full Version: FreeDV using Codec2 a low bit rate HF codec
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2
http://www.freedv.org/

We have an email group on Sourceforge:
freetel-codec2@lists.sourceforge.net

David Rowe a VK5 is developing codecs to rival the industry's best. A FREE codec
unlike the AMBE series used by P25, D-Star and DMR.

Download and compile the code from:
http://files.freedv.org/freedv/freedv-1.2.tar.xz
Or use the pre-compiled code from http://www.freedv.org/
I have it working here via an FT-450 radio, 80, 40, 20 and 6m.

Any interest?

Alan VK2ZIW
(13-05-2017, 09:02 PM)Hi Alan, so what are the Bene fits of this Digital audio and is it compatible with Dstar or P25 or DMR? I have no idea so trying to understand how it would be helpful. 73 Mike VK3XZvk2ziw Wrote: [ -> ]http://www.freedv.org/

We have an email group on Sourceforge:
freetel-codec2@lists.sourceforge.net

David Rowe a VK5 is developing codecs to rival the industry's best. A FREE codec
unlike the AMBE series used by P25, D-Star and DMR.

Download and compile the code from:
http://files.freedv.org/freedv/freedv-1.2.tar.xz
Or use the pre-compiled code from  http://www.freedv.org/
I have it working here via an FT-450 radio, 80, 40, 20 and 6m.

Any interest?

Alan VK2ZIW
Codec2 benefit:
1) Very low bit rate or data rate.
2) Open Source, we can implement it at NO COST.
      Not the $50 (or so) for the codec in P25, D-Star, Fusion and DMR.
     This means we are not limited to any proprietary hardware.

Compatibility:
   It was never meant to be compatible. AMBE as used in P25, DMR etc. require 9600BPS data rate.
   Codec2 is experimental with bit rates down at 1400 bits/sec and now lately 700 bits/sec.

How it works:
   Our human voice starts with our vocal cords then our mouth, tongue, lips and nose shapes. And
    added wind noise.
    Analyse this, send as digital data, recreate. Do this with the minimum data rate still maintaining intelligibility.

At 700bits/sec this can work well in our SSB 2.7KHz bandwidth at signal-to-noise ratios at the point where we humans
have difficulty.

If you've experienced P25 or DMR, the clarity can be much better (though it has an "accent") than regular noisy FM.

The services use digital as it can be encrypted. - Very important.
Our Emergency teams assisting them also need communications secrecy.
We are not allowed encryption. A codec is not encryption but you'd better be a good programmer if
we change say the bit order.

Alan VK2ZIW

VK2HRX

"We are not allowed encryption"

WICEN is.
(04-07-2017, 10:52 AM)vk3xz Wrote: [ -> ]
(13-05-2017, 09:02 PM)Hi Alan, so what are the Bene fits of this Digital audio and is it compatible with Dstar or P25 or DMR? I have no idea so trying to understand how it would be helpful. 73 Mike VK3XZvk2ziw Wrote: [ -> ]http://www.freedv.org/

We have an email group on Sourceforge:
freetel-codec2@lists.sourceforge.net

David Rowe a VK5 is developing codecs to rival the industry's best. A FREE codec
unlike the AMBE series used by P25, D-Star and DMR.

Download and compile the code from:
http://files.freedv.org/freedv/freedv-1.2.tar.xz
Or use the pre-compiled code from  http://www.freedv.org/
I have it working here via an FT-450 radio, 80, 40, 20 and 6m.

Any interest?

Alan VK2ZIW

HI Alan,
Thank you I will give it a try. Is there a Frequency normally used on the various bands to locate Hams using this method? Thanks and 73 Mike VK3XZ
Hi again all,

HF frequencies, yes, 7177KHZ LSB and 14236KHz USB.

We are a year on, FreeDV with Codec2 mode 700D is stable.

Would you guys consider a Parrot repeater using the 700D mode.
This would allow testing without a Full Duplex radio system including a duplexer.
Just one transceiver, antenna and computer. "Simples".

FreeDV would allow "secure" voice comms over HF, "secure" in that very few people could decode it.

All I'm asking is, is there any interest? WICEN etc..

Then, on what bands, HF or VHF?

Remember, when there's nobody on the other end of, say a 6m repeater, one can trigger it but I'd
like to hear my audio, not just the "kerchunk" and the CW ident.
That's why "parrot" enthuses me.

Alan VK2ZIW

VK3PR

A parrot repeater on 40m would be great. Setting it up would be well beyond me.
(04-06-2018, 03:40 PM)VK3PR Wrote: [ -> ]A parrot repeater on 40m would be great. Setting it up would be well beyond me.

I set up the FreeDV GUI software from freedv.org on Win7/64 bit yesterday, using my FT8-configured hardware with an extra headset, tried 7.177 & heard nothing, tried 14.150 - again nothing happening, but on 14.153 heard DV transmissions. Couldn't decode anything worthwhile via the GUI but recorded it to file and emailed that to David Rowe.He has responded confirming that it was DV 1600 and that he did get a decode. Maybe my receive levels need reviewing a bit - yet to be tried.  Had a listen on 14.236 at one stage and it was being used for SSB rather than DV at the time.
 
I was all set to try again at 0500Z today on 14.153, gear left running while I went off to a dental appointment. The dentist took longer than expected so arrived back home too late at 0545Z, so will probably try at 0500Z tomorrow, etc.

A FreeDV parrot would be useful in these early stages when trying to get gear functional but whether it should be on 40M or 20M - where more of VK and ZL can access it - remains to be seen. Of course, nothing will beat a real human on the other end as you flip between analogue and digital while trying settings, trying to get it right.

73 Doug - not quite yet a FreeDV convert  Wink
I succeeded in making a FreeDV QSO with VK3RV this afternoon on 14.153 at 0500Z although the receive audio was 'warbly'.  He answered me on my very first call so at least confirmed the transmit side works too.  The audio levels still need to be refined but it was a good starting point.

Peter VK3RV has skeds on that frequency and time at least most week days so is a good reference point to get your gear up and running on this mode. He also tries at 0600Z if he can't make the 0500 timeslot.
We can now do a repeater. A parrot repeater.

I've been battling with C code and with David Rowe's freebeacon.c code, created a repeater.

Because on HF, full duplex is out of the question, it's a parrot repeater or, store and retransmit.

It's running at the moment on 7190KHz LSB and needs a "trigger" string of VK2ZIW in your text.
Running on a Pi, a Banana Pi original with an SSD on it's SATA port. Fedora 28
Use FreeDV mode 700D

Alan VK2ZIW
Hi all,
Parrot repeater is UP. Mode 700D

I've done it, HOWZAT, all in the C code, no more calling scripts.

7177KHz with a trigger string of my callsign. Either upper or lower case.
Put my callsign anywhere in the text string on the FreeDV app.

Now running 24x7.

Alan VK2ZIW
After the Sunday broadcast:
The parrot repeater transmitted these:
I was at Dural but nowhere near a receiver....
I've played one of the longer files, a good QSO!!

-rw-r--r-- 1 alanb tims 256 Jul 29 11:23 done/2018-07-29-11:22:59_to_radio.700C
-rw-r--r-- 1 alanb tims 896 Jul 29 11:24 done/2018-07-29-11:23:51_to_radio.700C
-rw-r--r-- 1 alanb tims 1520 Jul 29 11:25 done/2018-07-29-11:25:16_to_radio.700C
-rw-r--r-- 1 alanb tims 3552 Jul 29 11:27 done/2018-07-29-11:26:12_to_radio.700C
-rw-r--r-- 1 alanb tims 832 Jul 29 11:29 done/2018-07-29-11:29:01_to_radio.700C
-rw-r--r-- 1 alanb tims 3632 Jul 29 11:31 done/2018-07-29-11:29:44_to_radio.700C
-rw-r--r-- 1 alanb tims 384 Jul 29 11:34 done/2018-07-29-11:34:08_to_radio.700C
-rw-r--r-- 1 alanb tims 2512 Jul 29 11:35 done/2018-07-29-11:34:42_to_radio.700C
-rw-r--r-- 1 alanb tims 4096 Jul 29 11:36 done/2018-07-29-11:35:59_to_radio.700C
-rw-r--r-- 1 alanb tims 0 Jul 29 11:52 done/2018-07-29-11:52:17_to_radio.700C
-rw-r--r-- 1 alanb tims 496 Jul 29 16:54 done/2018-07-29-16:54:14_to_radio.700C
-rw-r--r-- 1 alanb tims 464 Jul 29 16:54 done/2018-07-29-16:54:42_to_radio.700C
-rw-r--r-- 1 alanb tims 1120 Jul 29 16:55 done/2018-07-29-16:55:05_to_radio.700C
-rw-r--r-- 1 alanb tims 512 Jul 29 16:56 done/2018-07-29-16:56:19_to_radio.700C
-rw-r--r-- 1 alanb tims 544 Jul 29 16:57 done/2018-07-29-16:57:15_to_radio.700C
-rw-r--r-- 1 alanb tims 192 Jul 29 20:17 done/2018-07-29-20:17:09_to_radio.700C
-rw-r--r-- 1 alanb tims 576 Jul 29 20:17 done/2018-07-29-20:17:38_to_radio.700C
-rw-r--r-- 1 alanb tims 768 Jul 29 20:18 done/2018-07-29-20:17:58_to_radio.700C
-rw-r--r-- 1 alanb tims 480 Jul 29 20:19 done/2018-07-29-20:19:35_to_radio.700C
-rw-r--r-- 1 alanb tims 672 Jul 29 20:20 done/2018-07-29-20:19:54_to_radio.700C
-rw-r--r-- 1 alanb tims 288 Jul 29 20:30 done/2018-07-29-20:30:21_to_radio.700C
-rw-r--r-- 1 alanb tims 784 Jul 29 20:34 done/2018-07-29-20:34:22_to_radio.700C

Monday, I'm at a funeral...
Alan VK2ZIW
Having fixed the garbled audio transmission and adding the re-sending of the text string,
the repeater on 7177KHz LSB here QF56hg has been tested running well by Hil VK2AZ.
Put my callsign in the sent text string, this triggers repeat.

Alan VK2ZIW
Frequency is now 14.153 MHz USB

Perhaps I should have notified you all here as well as the SauceOnTheBarbie group.

It's running 24x7

I can shift the frequency to anything the FT-450D can do, antennas: 80, 40, 20 and 6m.

Alan VK2ZIW
March 22 lightning struck the tree next door!!

Dead FT-450D so for now the Codec2 repeater is off.

The Pi survived.
I've ordered an ozQRP MST3-SSB for the job.

Alan VK2ZIW
Hi all,
IC-910H now fixed, all bands, 2m, 70cm, 23cm.

New mode 2020 is out there now with a Windows exe on the website.
(No MAC as yet) and my compiles of the Linux source code has only worked
my Pi version running (Armbian on a Banana Pi M2 Berry)

Experiments with my "parrot repeater" and or SmallDV with this new mode have
not worked yet.

The ozQRP MST3 kit is built and working with the CDV VFO.

Alan VK2ZIW
A few weeks ago I received an Odroid N2 for Mode 2020 testing.
Still not enough go for mode 2020.
We need some SIMD programmers please?

So I'm back to code development of the parrot repeater.
Just before Codec2's move from Sourceforge to Github, somewhere between SVN 4010 and 4121
the FIFO code was changed. From this point "parrot" won't get sync, only occasionally.
Yes, I'm aware of the function name change eg. fifo_read => codec2_fifo_read
My testing is with mode 700D.

What you may have notice, David's SM1000 appliance has now a code upgrade to do mode 700D.

The new mode 2020 runs on Intel 64 bit systems both Windows and Linux but not MAC as yet so at this
stage my "parrot" repeater will concentrate on mode 700D.

Alan VK2ZIW
Hi Alan

Greetings from the UK. Heard from Matt that there was a group over there playing with smalldv - presume that is you guys? During the lockdown, we thought it would be a fun club project to keep members amused. Currently working fine 700D/700C/1600/800XA but was having a play with other modes. Realise 2020 will not run on Pi 3 but been trying Pi 4. Seems to run. If I try the 2020 LPCNet test (cat wia.wav | lpcnet_enc | ofdm_mod | ofdm_demod | lpcnet_dec | aplay) it runs and the audio sounds good but with a few underruns. CPU loading on the 4 running 2020 from fdvcore shows a CPU loading of 13% - not measured it properly yet for the above LPCNet test. Might be still not enough oomph but looks promising. Still playing this end but wondered if you guys had made any progress?

Best regards
Dave
G8KBB

http://warc.org.uk/?page_id=123372
Hi Dave,
I've now a RPi4B and have loaded HamVOIP for the RPi-2-3-4 only to discover it's 32bit.
The Codec2 library compiles but I'm yet to do anything with it. I have a stereo in and out
USB audio device, the Creative SoundBlaster Play 3! as these are available worldwide.
A stereo audio device should allow two separate paths for Rx and Tx for "smalldv".

The 2020 mode requires CPU hardware acceleration, for ARM this is "neon".
From the Intel i3 up (not Celeron) the AVX/AVX2 FPU extensions are 256bit.
This allows eight sixteen bit maths operations simultaneously.
"neon" is only 128bit.
88 (well we can drink in this lockdown)
Alan VK2ZIW
Hi Alan

Yes, happily working my way through stock of wine - including Australian :-)

Sent a little more info in the other message. If I try the 2020 LPCNet test (cat wia.wav | lpcnet_enc | ofdm_mod | ofdm_demod | lpcnet_dec | aplay) it runs and the audio sounds good but with a few underruns. If I split it ( cat wia.wav | lpcnet_enc | ofdm_mod > file ) takes 35%, and then the rest (ofdm_demod | lpcnet_dec | aplay) took 65 (ish I think from memory).
Compiling codec2 with LPCNet extensions for 2020 is pretty straightforward - I put a copy on the WARC website on the smalldv page.

At this end, Dave 'BCU is playing with the hardware, I'm doing the software.

Dave
G8KBB
Pages: 1 2