Re: [Freetel-codec2] FreeDV is hiring a DSP engineer

[glen english LIST]

Hi Mooneer

wow that list of requirements and resposibilities  reads like a 
USD$250,000/year job.  or more.


Perhaps it can be done by someone (a volunteer)  willing to manage a few 
volunteer contractors ? I'm happy to manage and coordinate as a volunteer.



On 9/02/2024 10:19 am, Mooneer Salem wrote:

Hi all,

The FreeDV project is looking to hire a DSP engineer! This person will 
be responsible for helping to develop the next generation of open 
source HF digital voice. Funding for this opportunity will be provided 
by the ARDC grant awarded to the project (thank you again!)


More information about this opportunity (including how to apply) can 
be found at http://bit.ly/freedv-dsp-engineer-job-listing.


Thanks,

-Mooneer K6AQ


___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Updating the libcodec2 binding in FFmpeg, fixed-point decoding

[glen english LIST]

It's an interesting idea. For real time work, taking it to the extreme, 
putting a bit of latency in there-


the encoder and decoder pair, could over time , generate shortcut tokens 
/ lookup  for whole words


IE encoder continuously stores the encoded words. (it doesnt know the 
word, only  the sequence ) and puts marker tags on them, and transmits 
the marker tag with the encoded sequence


at the other end, the decoder continously stores the sequences and the 
associated tag.


WHen the encoder finds a fitting sequence that's already been encoded 
and a close fit to work already down,  it only sends the token tag to 
the decoder to look up and reproduce.


IE a form of dynamic lookup table. There must be a name for doing this 
in computer science ???


glen


On 29/12/2023 11:54 pm, Tomas Härdin wrote:

fre 2023-12-29 klockan 10:44 +1100 skrev glen english LIST:

Hi Tomas
"codec2 + zip makes for even smaller files"

implying that there is still redundancy to remove in the codec2
encoded voice files 
That sounds like low hanging fruit to pick.

That is not surprising. In speech you can lower the signal rate by
talking more slowly. This and repetition happens often over the air.
Silence doesn't require more bits to encode than the length of the
silence.

I think Dave posted on his blog a while back that the effective data
rate of human speech is on the order of 1-10 bit/s or so

As for it being low-hanging, variable bitrate may be something to
consider for future modes. If there is silence, perhaps it is enough to
just transmit a carrier? If 14 bits are enough for a frame rather then
28 because reasons, then maybe Manchester code those bits. Or use a
lower baudrate for those bits, but I'm not sure how that interacts with
modem stuff. Oh and that would require a breaking change to the .c2
file format, unless you pad with zeroes

For files this isn't a huge issue I think. Might be fun to cram an
entire 10 hour audio book onto a floppy though

/Tomas


___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Updating the libcodec2 binding in FFmpeg, fixed-point decoding

[glen english LIST]

Hi Tomas
"codec2 + zip makes for even smaller files"

implying that there is still redundancy to remove in the codec2 encoded voice 
files 
That sounds like low hanging fruit to pick.

I guess though that is across a large number of frames (an audio book), where 
there may be redundancy / repeats of common codewords.  IEnot just a single 
frame. speech pauses , and for the same speaker, phonemes  and other speech 
components.

Interesting !

On 28/12/2023 9:09 pm, Tomas Härdin wrote:

codec2 + zip makes for even smaller files



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Codec 2 Algorithm description doc

[glen english LIST]

Hi Sebastien

I agree with your spirit and comment. I agree the current codebase can 
be a little difficult to follow !


I think the requirement and definition of "reference C code" requires 
that there is no optimisation, and no machine specific instructions , no 
unrelated or unused code etc. Is this your driving point ?


regards

glen english


On 14/12/2023 8:41 pm, Sebastien F4GRX wrote:

Hello,

I agree reference code is important as a known-good base to help test 
the compliance of third party implementations, but code NEVER has been 
equivalent to a specification.


Code is usually complex, and specific to a particular kind of 
machines. it uses shortcuts, optimizations, and is linked to its 
implementation language. it's almost art. It is unusable to build a 
hardware version of the codec. So, it's not a spec.


A formal spec is more important, as it describes the pure algorithmic 
structure of the codec in a concise, synthetic, yet practical way. 
Writing a spec is a also a way to make sure we know what the code is 
doing, without having to know how to code. And to make sure no detail 
is left out.


For example, there is a difference between writing "energy is 
quantified linearly by selecting codes in this lookup tables for these 
values of energies" and having to browse dozens of files to find which 
encoding table and routine are used for a specific mode.


I know because was there before, I tried as a third party developer, 
to document codec2 algorithms from code, without previous intimate 
knowledge of the algorithms (I have already shared my doc).


I found it very difficult to do, not because codec2 is complex, but 
because the current codec2 code base is quite large, comes with lots 
of unused and/or unrelated routines, and as such is very hard to 
follow. This is normal.


I am very glad that a specification is finally available. I have not 
yet been able to read it yet as github told me the pdf was corrupted, 
but I hope to get it soon, read it, help improve it, and use it to 
finally build a fixed point version.


Thank you very much for this document,

Sebastien



Le 13/12/2023 à 23:35, glen english LIST a écrit :

I support reference C code

For a developer who needs to implement an algorthm on their platform, 
the reference C code (RCC) provides ability to numerically  test 
bench their platform code (essential) , and provides a well broken 
down algorithm method in the form of C code (essential) .


In my opinion, an implementation developer does not need to 
understand how and why , just what has to be done.


I'm starting to implement Codec2 on a RISC-V soft micro on an FPGA 
with custom instructions. I started with the RCC, and identify 
regions that I can generate FPGA acceleration in the fabric , and I 
test bench it against vectors produced by the RCC.


(While I have a general interest in the inner  workings)   - I don't 
need understand David's excellent Codec2 math inclusive 
documentation, I just need to implement my numerically accurate 
version of the RCC


-glen VK1XX

On 14/12/2023 9:24 am, david wrote:

Hi Greg,

Thanks for your comments, good to get a perspective from a non DSP
person.  Some thoughts:

1/ The modes are not interoperable.

2/ The "stable" question is a good one.  I consider codec2 unstable, as
I am dissatisfied with performance and enjoy experimenting. Having
said that, the youngest mode is now 7 years old, and some haven't
changed in a decade.  I don't think we'd remove any existing modes that
had a user base > 1.




___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Codec 2 Algorithm description doc

[glen english LIST]

I support reference C code

For a developer who needs to implement an algorthm on their platform, 
the reference C code (RCC) provides ability to numerically  test bench 
their platform code (essential) , and provides a well broken down 
algorithm method in the form of C code (essential) .


In my opinion, an implementation developer does not need to understand 
how and why , just what has to be done.


I'm starting to implement Codec2 on a RISC-V soft micro on an FPGA with 
custom instructions. I started with the RCC, and identify regions that I 
can generate FPGA acceleration in the fabric , and I test bench it 
against vectors produced by the RCC.


(While I have a general interest in the inner  workings)   - I don't 
need understand David's excellent Codec2 math inclusive documentation, I 
just need to implement my numerically accurate version of the RCC


-glen VK1XX

On 14/12/2023 9:24 am, david wrote:

Hi Greg,

Thanks for your comments, good to get a perspective from a non DSP
person.  Some thoughts:

1/ The modes are not interoperable.

2/ The "stable" question is a good one.  I consider codec2 unstable, as
I am dissatisfied with performance and enjoy experimenting.  Having
said that, the youngest mode is now 7 years old, and some haven't
changed in a decade.  I don't think we'd remove any existing modes that
had a user base > 1.




___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Linux-preferred desktop FreeDV?

[glen english LIST]

Hi Adrian, cheers.

Yeah thinking that the most popular installed environment would be the 
one with the dependences well sorted, searching for this on google, I 
couldnt find a consistent answer.LOL I've always run XFCE  
seems to have the smallest installed footprint but of course that doesnt 
tell you much about CPU consumption and how well it is written.



On 23/11/2023 8:24 pm, Adrian Musceac wrote:
KDE Plasma 5 on Debian works well too. I think it's more like an issue 
of the distro packaging all the dependencies in a consistent manner.


Adrian


On 23 November 2023 00:35:11 UTC, Mooneer Salem  wrote:

Hi Glen,

I'm not sure it actually matters given FreeDV's use of wxWidgets.
While I usually use macOS, whenever I test on Linux I personally
just use the GNOME desktop environment and have no problems.
___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

[Freetel-codec2] Linux-preferred desktop FreeDV?

[glen english LIST]

On Linux Debian, you get the option of XFCE, gnome, KDE, cinnamon, MATE, 
LXDE, and more


I ask what is the most SUITED desktop environment for new installs to 
support the FreeDV GUI code ?


glen VK1XX



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] STM32 freeDV cost benchmarks

[glen english LIST]

Hi Mooneer

thanks for the comments. The FDIV on the ESP32-S3 is very slow I think 
80 cycles, but of course one should always use RECIPS  and a MUL. 
Integer div is a couple of cycles IIRC.


I'll start and see what jams it all up .

regards,

On 21/11/2023 4:01 pm, Mooneer Salem wrote:

Hi Glen,

I seem to remember that there was a previous project that implemented 
it on a FPGA, but I can't recall specifics at the moment. FWIW, on 
ezDV (ESP32-S3 but should perform similarly to the STM32F4 on the 
SM1000), 700D TX took something like ~90ms per 160ms frame while RX 
using real-only (vs. complex numbers) was something like 40ms per 
160ms frame. Unfortunately, I never did do a detailed accounting of 
what parts of encode and decode took the most time on the ESP32.


Thanks,



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

[Freetel-codec2] STM32 freeDV cost benchmarks

[glen english LIST]

Hi Mooneer

I am preparing to implement FreeDV on an FPGA.

Is there a location in the repo somewhere, or does anyone have , 
anecdotally a feeling of, the major workloads / major consumers of 
processor time on the STM32 FreeDV port.


regards
___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Codec2 in real time on STM32

[glen english LIST]

Hi Jan

Just adding to the other two very good replies

My experience with codec2 with STM32F4/F7 At O0, the performance may be 
down to 50% of -O2, and also, with O0, its likely the compiler is not 
exploiting any hardware speedups as with GCC there are some (invisible)  
ties between -O level, and the use of the hardware flags.


However ! I would certainly investigate why you have hardfaults. It may 
be that the processor is not keeping up with interrupt processing etc,


but I would suggest -  put some code in there for the hard fault handler 
so it stops and you can see why happend


best
glen
here are some examples:

this is an EXCELLENT primer :

https://www.freertos.org/Debugging-Hard-Faults-On-Cortex-M-Microcontrollers.html?_ga=2.258623882.1201655825.1700519821-15576747.1700519821

***and code I use :

void HardFault_Handler( void ) __attribute__( ( naked ) );


void prvGetRegistersFromStack( uint32_t *pulFaultStackAddress )
{
/* These are volatile to try and prevent the compiler/linker optimising them
away as the variables never actually get used.  If the debugger won't 
show the
values of the variables, make them global my moving their declaration 
outside

of this function. */
volatile uint32_t r0;
volatile uint32_t r1;
volatile uint32_t r2;
volatile uint32_t r3;
volatile uint32_t r12;
volatile uint32_t lr; /* Link register. */
volatile uint32_t pc; /* Program counter. */
volatile uint32_t psr;/* Program status register. */

    r0 = pulFaultStackAddress[ 0 ];
    r1 = pulFaultStackAddress[ 1 ];
    r2 = pulFaultStackAddress[ 2 ];
    r3 = pulFaultStackAddress[ 3 ];

    r12 = pulFaultStackAddress[ 4 ];
    lr = pulFaultStackAddress[ 5 ];
    pc = pulFaultStackAddress[ 6 ];
    psr = pulFaultStackAddress[ 7 ];

    /* When the following line is hit, the variables contain the 
register values. */

    for( ;; );
}

void HardFault_Handler(void)
{
    __asm volatile
    (
    " tst lr, #4    \n"
    " ite eq    \n"
    " mrseq r0, msp \n"
    " mrsne r0, psp \n"
    " ldr r1, [r0, #24] \n"
    " ldr r2, handler2_address_const    \n"
    " bx r2 \n"
    " handler2_address_const: .word prvGetRegistersFromStack    \n"
    );
}


On 21/11/2023 12:21 am, Jan Ropek wrote:


Hello,

my goal was to get Codec2 (encoding and decoding) working on STM32F446 
running on 180MHz. So, I created a new project for Nucleo with F446RE 
and added Codec2 libraries to it. I compile with GCC (using 
STM32CubeIDE - based on Eclipse).


I have GCC optimization turned off and encoding and decoding works, 
but the problem is that encoding one 40ms frame (320 B) takes about 47 
ms, so I am not able to use Codec2 in real-time. So, I am asking 
whether my requirements are unrealistic, or have I implemented Codec2 
incorrectly?


- I tried comparing CMSIS FFT and KISS FFT, it takes roughly the same 
time.


- When I turned on GCC optimization to O1 - encoding sped up to about 
15 ms, but during decoding, a HardFault always occurred.


Can you please suggest what I might be doing wrong? Many thanks!

Best regards, Jan.



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Starting an organization for FreeDV/Codec2 development

[glen english LIST]

Hi Mooneer

Great initiative. I started down a hardware road about 18 months ago- 
the idea was a retrofitable board to go in a radio or a microphone, or 
an appliance.


I though also of a device (appliance) that could be plugged directly 
into a USB sound card and be integrated into mic and speaker, and 
perhaps have BT etc or other itnerface to permit it being a data modem.


But the lack of parts 6 months in has stifled everything- its hard to 
buy anything.


I had considered that there needed to be enough horsepower to do the 
LPCnet modes, (which certainly sets a floor on the hardware) and there 
are a couple of spoke to that were interested in contributing.  The 
LPCnet modes , the need to support those, I wonder how widespread that 
thinking is for hardware ???


As you might know i am a professional design engineer, have Altium, FPGA 
packages, Xilinx Alliance Partner etc (which helps me buy some parts but 
not loads) .


I also have a modern Samsung SMT machine SM482 (32,000 chips per hour 
top speed). I think it would take about 15 seconds to load a SM1000 grin 
grin. setup time is something else though (a day).


-glen VK1XX


On 21/02/2022 9:07 am, Mooneer Salem wrote:

Hi all,

Over the past year or so, there have been several occasions where it 
would have been beneficial to have had an official organization 
steering FreeDV and Codec2 development. For instance, the ARDC does 
not currently offer grants to individuals 
, not to mention the recent 
discussion about Windows Defender 
 and what it would 
take to get FreeDV "trusted" by Microsoft. I'd like to brainstorm what 
such an organization would look like.


First off, what would the mission of such an organization be? This 
seems obvious at first glance ("promote the use of open source digital 
voice modes on amateur radio"), but what would that entail exactly? A 
few potential things off the top of my head:


* SM1000 replacement (faster hardware to support other FreeDV modes, 
Wi-Fi support for radios such as the IC-705, etc.)
* General software process improvements (code signing of binaries, 
hosted Debian/Fedora/etc. package repos, etc.)
* General advocacy (e.g. presence at more ham radio related events, 
Zoom/etc. presentations about HF digital voice)
* Offering grants to individual developers to develop Codec2 and 
FreeDV further (e.g. fixed-point ports, improved/more modes, etc.)


On first thought, it seems that a non-profit/501(c)3 structure would 
be most appropriate given the mission of such a potential organization 
and its dedication to open source, but I'm wondering if some other 
structure would fit better. Another possibility is structuring it more 
like a membership club (IIRC there was a digital voice club back in 
the FDMDV days but I forget offhand what it was).


Anyway, if you have any suggestions or comments, they would be most 
welcome.


Thanks,

-Mooneer K6AQ


___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2


--
Glen English
RF Communications and Electronics Engineer

CORTEX RF

Pacific Media Technologies Pty Ltd trading as Cortex RF

ABN 40 075 532 008

PO Box 5231 Lyneham ACT 2602, Australia.
au mobile : +61 (0)418 975077



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] 1600 mode and constellation points

[glen english LIST]

Hi Brian

WRT the pilot levels, - other OFDM HF modes use 3dB. That David's 
implementation is only 1dB pilot boost is excellent, and shows a high 
skilland best practice.


Have a look at the specification for another HF OFDM mode like DRM. 
There are many boosted pilots, Tthe power boost for DRM is 3dB, so in 
many respects, David's implementation is excellent for a pilot boost of 
only 1dB- very impressive.


I think the opinion from your QAM friends may illustrate they are  
il-informed about HF OFDM techniques.


Download ETSI ES 201 980  and take a look,  section 8.4 pilot cells. 3dB 
and walking pilots are used . DRM is in general transmitted up to 20kHz 
bandwidth and more pilots are required.


In DVBT- same power is used but the pilot data is repeated and the demod 
averages over multiple periods.


-glen english


On 25/08/2021 12:03 am, Brian Morrison wrote:

On Mon, 2021-08-23 at 10:43 +0930, david wrote:

The BPSK pilot is used for acquisition (e.g. estimating the initial
frequency offset) and frame sync.  More power is allocated to the
BPSK
pilot to make those two functions fast and reliable (an important
requirement for PTT digital voice).  As there is only one BPSK
carrier,
the overhead in terms of total power is about 1dB IIRC.

Yes, I thought acquisition might be a reason. The 1dB power overhead is
not much, but for some other systems using QAM a designer would be told
that 1dB is unacceptable.


The FreeDV 1600 waveform (the FDMDV modem) was an open source
implementation of an earlier waveform (see the History section on
http://freedv.org) that was known to work well for this application.
  So it was something of a starting point for FreeDV, and a stepping
stone to later waveforms.

There are Octave and C versions of the modem (
https://github.com/drowe67/codec2/blob/master/README_fdmdv.md), plus
various blog posts, if you would like to experiment with the modem or
inspect it's implementation.

OK, I now know where to look if I decide to have a play.

Thanks for your answer, very illuminating.




___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Questions about Codec2 embedded performance

[glen english LIST]

I think, based on my profressional experience with the F4 , L4, L, in 
battery environments -  you would be far better  to  use an F4 part for 
this.


-aggressively down clock it when the full clock speed is unnecessary
You can save a ton of power doing that.
The chip is very good at changing clock speeds.

- you can disable cache, disable flash prefresh all saves heaps of power

- you can use a switched mode instead of LDO for the core if you really 
need to.  The core runs on 1.2V or down to 1.1V

that saves power.

- downclock peripherals, aggressively gate their clocks.

My opinion is that the ONLY reason to use an L4 for this is if up need 
the thing to sleep doing NOTHING except waiting for an interrupt to wake 
up.


The F4 will run on 1.8V. forget 3.3V . logic tralsate asd required. 1.8V 
is the key to low power.


You might even use the F7 part and again, use the above methods to save 
power. The F4 happily RUNS at 2mA power consumption when unbusy.


You can go to stop for 90% of the time and consume just 200uA average.




On 5/14/2021 1:04 PM, Josh Lloyd via Freetel-codec2 wrote:

Glen

 > But, certainly it should compile and run and you would be able to
generate benchmarks for your implementation- that's really what you need
to do.

That's probably what I'll do, get an L4 and just test it, though knowing 
roughly if it'll work before I go through the effort of shipping a 
device is always a win in my opinion. I'm learning towards "it wont work 
without effort" though.


 > Exactly what power budget (milliwatts) do you have ?

I'm targeting less than 1 watt, probably around 500mW for the radio and 
processor and other peripheral, and making use of low power sleep modes 
to conserve energy in standby or while receiving.


 > Do you need to encode real time ?

Some latency is tolerable, but the intent would be real time - yes.

Josh



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2




___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Questions about Codec2 embedded performance

[glen english LIST]

Josh

L4 has the FPU , it IS an Cortex M4  core.
but there are many many low power optimizations internally , with 
respect to caching, memory bandwidth etc.  Per is specificed at 
1.25DMIPS/MHz for both F4 and L4.  Some of the  L4 low power facilities 
are NOT available when running at performance of 1.25DMIPS/MHZ.


I think that you may want to reevaluate your strategy.

But, certainly it should compile and run and you would be able to 
generate benchmarks for your implementation- that's really what you need 
to do.


Exactly what power budget (milliwatts) do you have ?
do you need to encode real time ?

Glen English
STM32L0, L1, L2, F0 , F4, F7  and Xilinx FPGA person.




On 5/14/2021 8:51 AM, Josh Lloyd via Freetel-codec2 wrote:

Hi Bruce, thanks for your reply.

 > Hardware floating point is requ



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Re: [Freetel-codec2] Questions about Codec2 embedded performance

[glen english LIST]

Josh

L4 has the FPU , it IS an Cortex M4  core.
but there are many many low power optimizations internally , with 
respect to caching, memory bandwidth etc.  Per is specificed at 
1.25DMIPS/MHz for both F4 and L4.  Some of the  L4 low power facilities 
are NOT available when running at performance of 1.25DMIPS/MHZ.


I think that you may want to reevaluate your strategy.

But, certainly it should compile and run and you would be able to 
generate benchmarks for your implementation- that's really what you need 
to do.


Exactly what power budget (milliwatts) do you have ?
do you need to encode real time ?

Glen English
STM32L0, L1, L2, F0 , F4, F7  and Xilinx FPGA person.




On 5/14/2021 8:51 AM, Josh Lloyd via Freetel-codec2 wrote:

Hi Bruce, thanks for your reply.

 > Hardware floating point is requ



___
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2