I have noticed by playing my Vorbis files in Winamp that the bitrate used for different passages within a regular VBR vorbis file does not vary much, much less than for a VBR MP3. Also, during ABX testing, within a single song, some passages become transparent at around q=2 or q=3 while others I can tell up to q=5. Some entire songs are transparent at q=3 while others aren't till q=6. These tests were done encoding with OggdropXPd with Libvorbis 1.1.0.

If they theory of VBR encoding and q settings instead of bitrates worked like it should, the whole song would become transparent at the same time. Basically what I think needs to be done is that VBR needs to be MORE variable, i.e. spend even more bits on complex passages and even less on simple ones at a given quality level.

BTW, while I'm posting, what setting in vorbis encoders based on libvorbis 1.1 is roughly equivalent in quality to -preset standard in LAME, according to the general consensus?

EDIT: If u encounter problems with Vorbis 1.1.0, u can try these

http://www.rarewares.org/files/ogg/oggdrop...7.11aoTuVb2.zip

http://www.rarewares.org/files/ogg/oggdrop...11megamixP3.zip

http://www.rarewares.org/files/ogg/oggdrop...1megamix2P3.zip


use one of these oggdrop and try to ABX again
anyway, auToV is very good at q=4

there's nothing wrong with vorbis VBR
its pure VBR, Vorbis can use any bitrate for any frame
unlike mp3 which can only use preseted bitrate for each frame(i.e. 32,64,128,160.192....)
Vorbis has a better VBR implementation than mp3,
besides, the m/s stereo of Vorbis is much more efficient

The Vorbis plugin for Winamp updates its bitrate display much more slowly than it's mp3 plugin. It is changing pretty greatly per frame. Play the file in foobar with the right settings and you will see how it changes frame per frame.

Official 1.1.0 isn't bad at all. In fact it is the recommended build here.

In my experience you'd definitely be safe using -q 6. I use -q 5 personally and can't ABX it.

sorry, I shouldn't say it that way
"no no no" is removed

Vorbis 1.1.0 will do better with some kind of "impulse" switch activated
to save myself from those tedious process,
I would rather use megamix which is also using Vorbis 1.1 (RC1)

is megamix similar to --r3mix?
its just a compile that activates all the long switches?

thanks

Nope, Megamix is a combination of different tunings, lika AoTuV, QK (and maybe Garf's too).
Check the vendor tag of a megamix encoded file...(which I didn't do now)

Megamix = aoTuV b2 + QKTune 3.2 + GTune 3 beta 2
Megamix II = 1.1 RC1 + QKTune 3.2 + GTune 3 beta 2

Megamix 2 is definitely better...can't ABX that at q4. Haven't tried AoTUV yet cause I can't ABX much more right now w/o killing me eardrums cause this stuff needs to be played at insane volumes to make the subtle differences audible.

Also, a note about the importance of ABX testing: I thought for sure I heard the differences in one song at q4 between the original and encoded, since I was ABXing w/ hidden results. But when I got the results afterword I actually got less than half of the trials right, indicating that there's definitely no way I heard a difference. Is there any real point in trying Megamix original or is the consensus that megamix 2 is clearly superior?

Now that my eardrums have recovered I ABXed AOTUV and it becomes transparent to me at q4 also. The thing is AOTUV provides the least annoying artifacts, I think, with no annoying artifacts that are obvious even without doing a side-by-side comparison all the way to q0.

I agree for sure with the people on this board who say don't tell us how you a given codec is not transparent until you've ABXed it, because again, I thought for sure I heard differences, but after 15 trials (Predetermined, did not look at results till after the 15th trial), I got 8/15, clearly showing that I couldnt tell especially since I was looking only at a difficult ~1.5 minute passage. The placebo effect is huge. Also, using Winabx, sometimes I would leave the player set to either A or B instead of X before clicking next trial, and not realizing that it was on A or B I would click next trial and think I heard a discernible difference in color indicating that X had switched, until I realized I was listening to A or B.

I think its up to you
megamix n megamix 2 looks quite the, the only difference
is auToV b2 in megamix n 1.1 RC1 in megamix2
the bitrate produced by megamix2 is always ~10 kbps larger than megamix



I feel that too, for < -q4, < 128kbps, i think auToV produce
the least artifact among all the Vorbis encoder, that's why
auToV manage to win the 128kbps test!

auToV is specially tuned to overcome the high frequency boost




That's why this board is cool compared to other no brain website
that uses FFT graph to tell the quality of an encoder......
the result is WMA is the best coz it has the closest "look" compared
to the original, and preserve the most fidelity


anyway its normal, at -q 4, 128kbps its impossible for any
codec to be transparent for
ppl that still have "young/flexible" eardrum using very good to moderate
stereo system.

since its already hard for you to tell the difference between
-q 4 and the original, i think you can safely encode eveything
to -q5 and if possible, -q 6

as for Vorbis < -q 6, lossy stereo mode is used IIRC
You might not be able to hear the difference at -q 5 at all even
though lossy stereo is used but to be on the safe side,
use -q 6 and above! I'm afriad it might screw up when
encoding surround music?

What do you mean by right settings? Does Foobar show "average" frame changes in eg. 'update every 16 frames'? What would be the ideal setting?