[Asterisk-Dev] writing a GPL G.729?

[Steve Kann]

Steve Underwood wrote:

> Steve Kann wrote:
>
>> Race Vanderdecken wrote:
>>
>>> The G.729 people/corporations/researchers and such spent a bunch of 
>>> time
>>> figuring it out so they deserve to have the right to charge for it.
>>>
>>> These are the rules of the game. Don't like it? Write it yourself and
>>> then give it away. But don't ask others for the shirt off their back
>>> because you are left out in the cold Mr. Grasshopper.
>>>  
>>>
>> I don't know anything about Mr. Grasshopper's temperatures, but I 
>> would say that you don't need to write it yourself..
>>
>> Just use speex, and encourage hardphone vendors to do so as well.
>>
>> It is the only high-quality, patent-free codec out there. If people 
>> would invest half the effort they are putting into whining about 
>> G729's patents into supporting patent-free alternatives, we'd all 
>> have less headaches to worry about.
>
>
> I fully agree that promoting Speex is the right thing. I recently 
> heard that Vorbis has gained real traction in the gaming world, 
> because it avoids the licence fees associated with any other music 
> compression. It seems, therefore, that positioned correctly it might 
> be possible to see Speex gain some traction in voice compression too. 
> However, right now the fixed point implementation of Speex still seems 
> a bit weak, and I don't think any optimised versions for things like 
> 54x, 2181, etc. exist. Without that, it won't get very far in VoIP.

I think people are working on this;  There's certainly posts all the 
time about people looking to port the codec to different DSP 
implementations -- someone from hp seems to be porting to TI C5x now.  
He reported earlier:

I have the encoder and decoder running now and have verified that the 
> encoder is bit-exact wrt to the fixed-point code running on x86 for the 
> same 30-second audio sample.  Encode and decode together run in 
> real-time for 8KHz data, complexity=3, on 120MHz C5509 when code and 
> data are all in on-chip SRAM.  I have not tested the wideband codec yet.

-SteveK