[asterisk-dev] iLBC codec

[Steve Underwood]

On 07/13/2011 04:41 PM, François Delawarde wrote:
> On Wed, 2011-07-13 at 01:49 +0800, Steve Underwood wrote:
>> iLBC actually sounds a little worse than G.729, despite running at
>> twice
>> the bit rate.
> Just out of curiosity, do you mean the graph and statements from their
> web is wrong when they compare it to G.729?
> http://www.ilbcfreeware.org/
>
> François.
>
You'll notice in that graph that G.723.1 scores a little lower than 
G.729 and iLBC, and iLBC scores exactly the same as G.729. Most stuff on 
the web shows a higher MOS for G.723.1 than for G.729, which it totally 
bogus. These are subjective scores, so you generally get the result you 
are looking for. :-)

Perhaps you are looking at the degradation of these codecs with rising 
packet loss. Those curves are certainly bogus. iLBC is about twice the 
bit rate of G.729, so you can send G.729 with some redundancy, end up 
with the bit rate of iLBC, and have a better overall result.

Steve