Re: [codec] comparitive quality testing

Steve,

I was actually not 100% certain which flavor of G.729 should be selected. I
had exactly the same reasoning as you that G.729 is better quality, but
G.729A is more commonly used. Since my primary goal is to compare with
something that most people find acceptable quality for a voice call, testing
against G.729A is probably more appropriate.
_____________
Roman Shpount

On Fri, Apr 15, 2011 at 1:14 PM, Steve Underwood <steveu@coppice.org> wrote:

> Hi Roman,
>
>
> On 04/16/2011 01:01 AM, Roman Shpount wrote:
>
>> David,
>>
>> We should compare with standard G.729 (not A or B) at 8 kbit/s, 0 and 5%
>> packet loss, clean speech and speech mixed with at least one noise sample.
>> We can use ITU-T audio library for both speech and noise samples.
>> _____________
>> Roman Shpount
>>
> I fully agree that G.729 should be tested without the B annex, but I think
> your other choice is questionable. G.729 may sound distinctly better than
> G.729A, but G.729A is the dominant codec in the real world. Basic G.729 is
> actually quite rare. Are you trying to test against what people are familiar
> with, or what a pretty good current generation 8kbps codec can achieve?
>
> Steve
>
> _______________________________________________
> codec mailing list
> codec@ietf.org
> https://www.ietf.org/mailman/listinfo/codec
>