Well, since Ogg Vorbis is free, does not violate patents etc etc etc AND it performs similar to even MPC (I see that in table) and it even beats mpc in SOME samples - Vorbis could not be not recommended for archival mp3 is history - you can not sqeeze too much from it, but Ogg Vorbis has large tuning potential So Vorbis beats mp3 and mpc in lower bitrates, beats in low-mid ranges (look at 128kbps test) and it is ~equal in mid-high bitrates. So I know which format is most universal

According to Roberto's conclusions, vorbis didn't beat mpc. And according to my test, vorbis isn't ~equal to mpc. At least not at the same bitrate range. Vorbis is clearly inferior on most samples I've already tested (it means the 10 present in this test, and some other I've tested occasionally those two last years).

And I think it's better to take into consideration AAC format: it's better than all other solutions at low bitrate (let's see the results of 32 kbps test, but at 64 kbps, old HE-AAC had a serious margin), competitive at mid bitrate, and has also great performances at mid/high bitrate.

Roberto conclusions are simply wrong. Ogg Vorbis had more point and it is winner, period. Roberto is well known AAC zealot who don`t like Vorbis, I don`t care his conclusions, results are clear - Vorbis is SLIGHTLY better in 128kbps test.

According to your test Vorbis IS (only) slightly worse than MPC - in points (well, picture is gone, can`t give numbers) and is (slightly?) better than overtuned mp3.

I think that it is not better to take into consideration AAC format - free implementations are poor, not-free are not free - I don`t want to pay for this since I have competentive free alternative - maybe (HE-)AAC is slightly better, but not so much better to pay for it! Pay for encoder, pay more for NOKIA cellular phone (yes, price for HE-AAC is simply included in price of mobile) etc.

Also - Vorbis is simply sufficent for most people for most of bitrate ranges (mp3 fails for lower bitrates). If AAC/MPC is sooooo super, and Vorbis SUX, why Rio Karma, Irivier etc. and many game developers choose Vorbis (even outdated encoder, 64kbps)? Nokia is exception, one of not too many. For internet radios and music in games market is divided mostly between mp3 and Vorbis...

Also Vorbis has so much room for tuning - 1 year ago noone thought about results like in this test, in next couple of... years maybe months we`ll see more improvement.

PS. I wonder if HE-AAC integer decoder takes less or more CPU power / RAM than Vorbis one?

I vote vorbis was a loser there. Statistically it has a bit less confidence, but anyway...
"Much to learn you still have" (Yoda)
BTW, it would be very interesting to see graph with statistical margins for this test...

Well, IIRC Garf told >1 year ago that in fact all codecs are almost same Of course this test (128kbps) was poor performed - IIRC it wasn`t blind test AND bitrates were not equal. I vote for such testing:

- set level of test (e.g. 128kbps)
- for VBR/ABR, encode files many times until reach averange 128kbps FOR EACH FILE!!!

then test... in BLIND TEST

Could a moderator remove Celsus post about "poor" conducted and "wrong" conclusions of Roberto's test from this topic. Sounds like trolling to my ears (Vorbis had more point than MPC at 128 kbps => therefore it's the winner ; Vorbis had less point than MPC at 175 kbps => therefore it's the ~same quality).

Thanks.

Hey, I wrote "According to your test Vorbis IS (only) slightly worse than MPC - in points (well, picture is gone, can`t give numbers) and is (slightly?) better than overtuned mp3." So I told that it is worse - a bit...

EDIT: and people can read and see such things and have their opinion, so no need to delete post...

You probably miss this important one:
http://www.hydrogenaudio.org/forums/index....ndpost&p=225161

"Vorbis perform similar to MPC". I can't see it in my results. And the difference is much greater than the existing difference wetween mpc and vorbis at 128 kbps, which is, according to your own words, the proof of the superiority of vorbis ("more point = winner").

EDIT: incorrect link

Well, it was in "conclusion type of Roberto", in my one Vorbis was worse.