[Asterisk-Users] Open G.729A codec

[Mark Spencer]

> This Windows binary is probably fairly easy to convert for someone with
> sufficient skills. It's a simple library, COFF format. It's probably
> sufficient to split it into .o files (using ar), then convert the .o
> files (using objcopy --target=elf32-i386, objcopy from cygwin has both
> elf32 and coff formats, so it's useful for that), and assemble the
> resulting elf32 .a library (again, using ar). What remains to be taken
> care of are mostly underscores in function/variable names.

It's a little more complex than that.  Remember the Windows one is
single-channel only.  It's not reentrant and thus totally useless for
Asterisk unless you only need one channel.

> Otherwise, this process should work and one should be able to create a
> working Linux library (along with an asterisk codec).

Which could not be distributed without violating GPL, nevermind Voicages
licenses.  See:
http://www.fsf.org/licenses/gpl-faq.html#GPLModuleLicense
http://www.fsf.org/licenses/gpl-faq.html#MoneyGuzzlerInc

> Just remember that this is for non-commercial, personal usage only, as
> the license clearly states. Also, one must not reverse-engineer the
> code, which the license prohibits.

A requirement which you cannot apply to GPL'd code (unless you were the
copyright holder as Digium is and thus able to make such exceptions).

> Then I gave it some thought and couldn't really find a reason to do so
> much work on non-free code while there was speex almost ready to be
> used.

Speex is really a great thing, but G.729 is the unfortunate standard for
communicating with most (proprietary) SIP/H323 devices.  If ATA 186's
could talk SpeeX this wouldn't be a problem.  Trying to get the Windows
G.729 code ported to run with Asterisk is definitely barking up the wrong
tree though, for both technical and legal reasons.

> I think it is rather sad (not to say ridiculous) for a company to guard
> a piece of code this small with such monstrous licensing schemes.

Amen!  Especially when essentially the code is entirely available via the
ITU web site for a nominal fee!  <not speaking as Digium
here>Unfortunately, companies like Voiceage/Sipro who purely are IP
licensors apparently have nothing better to do than to try to come up with
such rube-goldberg schemes for copy protection, even while demonstrating a
remarkable lack of disregard for the actual quality of their
implementation (especially in the early days).

> Couldn't agree more. The G.729 codec is so unDigium-like... don't buy
> it is my recommendation.

I don't think anybody buys G.729 just to have it.  They buy it because
they *have* to have it.  And we sell it because they *have* to have it.  I
think eventually we'll be able to come up with a better (but not, for the
near future, open) G.729 solution from us.

Mark