I have noticed when using the alt-preset standard with lame 3.90.3, that the quality setting is rated at 2, which is equivalent to the -h-switch. Would it improve quality to override this with the -q 0-switch?

I wonder because it has been said on numerous occasions that one "should not mess with the alt-presets" and that only the one that has created them fully understands them.


Erlend

Or you could just be safe and use --alt-preset extreme

That was not the point: Going to extreme would change the bitrate, while specifying -q 0 would change the algorithms used, not the bitrate. Am I totally off here, or is there some sense to what I am saying?

I'm 99% sure that going q0 would put the bitrate WAY up, but try it yourself and see what happens.

I'm pretty sure the -h switch uses the better algorithm quality selection. Same goes for the newer (but not the better) lame version 3.93.1. The -q 0 switch is messed up there. Whatever, trust lame 3.90.3 & --aps :-)

I used to use -V 0 -b 128, with 3.93.1, before I knew about the alt-presets, and it always worked fine. note: -V 0 is the same as (-v -h -q 0).

edit: I just spelled something wrong and fixed it.

Tried the following:

File:

R.E.M. - New Adventures in Hi-Fi - 06 - That's What Keeps Me Down.wav (77 147 996 bytes)

Using --alt-preset standard it gave this file

R.E.M. - New Adventures in Hi-Fi - 06 - That's What Keeps Me Down.mp3 (12 274 191 bytes)

Using --alt-preset standard -q 0 gave this file

R.E.M. - New Adventures in Hi-Fi - 06 - That's What Keeps Me Down.mp3 (12 254 269 bytes)


So as you can see, the no -q 0 was larger (but not much).


I'm trusting the --alt-presets, I'm just wondering why --alt-preset standard would be better than --alt-preset standard -q 0?

According to the description in the .htmls enclosed in Dibrom's own compilation (and every other lame-compilation to my knowledge) it says:

-q 0..9 algorithm quality selection
Bitrate is of course the main influence on quality. The higher the bitrate, the higher the quality. But for a given bitrate, we have a choice of algorithms to determine the best scalefactors and huffman encoding (noise shaping).

-q 0: use slowest & best possible version of all algorithms. -q 0 and -q 1 are slow and may not produce significantly higher quality.

-q 2: recommended. Same as -h.

-q 5: default value. Good speed, reasonable quality.

-q 7: same as -f. Very fast, ok quality. (psycho acoustics are used for pre-echo & M/S, but no noise shaping is done.

-q 9: disables almost all algorithms including psy-model. poor quality.

AFAIK, -q0 is experimental. Check out Suffer.flac on ff123's Samples page; it's a good file for testing -q0 vs. -q2 (h).

Experimental regarding quality or speed? I mean, it's much slower than -h, but if the quality is slightly better, why not use it? I don't know much about this, but if some developer could explain why, I would be happy.

I'm resurrecting this old thread because the same question has come up for me.

Are the -q0 and -q1 switches, in fact, "messed up" and "experimental"?

I'd like to use the best quality algorithm for some speech files that I'm encoding at very low bit-rates. Is there any drawback to using the -q0 switch in conjunction with -V8 --vbr-new, using LAME 3.96.1?

Thanks in advance for any help,

~esa

Gabriel commented on the -q0/-q1 bug recently here:

http://www.hydrogenaudio.org/forums/index....ic=18091&st=25#

I'm not sure exactly what the bug was though.

to clear up some misinformation in this thread, first of all the -q settings don't/shouldn't affect bitrate at all. Second of all, no one has ever confirmed that q values higher than 2 (which is selected by the presets or by using -h) actually do anything to improve quality. They certainly slow down the encode process, but I wouldn't use em. All the preset tuning happened with the default -q 2.

In LAME 3.95 and newer, they remapped -q2 to -q3 because a new super-slow option was added at -q0... recently some bugs were found as well.

In conclusion... use -q2 it's faster and safer.

Thanks, guys!



~esa

so is it better to use -q2 than the -q3 that is defined by the
presets in 3.96.1?

and what potential harm might "messing up the presets" do?

.veikko

no, -q2 in 3.90.3 is the SAME as -q3 in the newer versions. Using -q2 now would be like using -q1 in 3.90.3, ie not recommended.

ah, ok thanks.

so i'll stick with the combination of 3.96.1 / -V0

(only about 50 albums to go, glad that I didn't find out now that I need to change
the settings of lame )

.veikko

for best quality you shouldn't be messing with the -q switch (or with additional switches in general). Just stick with the --alt presets and you'll be fine.

LAME 3.96.1 uses -q3, which is the same as -q2 in LAME 3.90.3

don't you think that -V0 is overkill? Can you hear the difference between -V2 and -V0? I doubt it...IMHO everything lower than -V2 is a waste of space...

Offtopic: That song is called "Leave"

Well it certainy might be overkill because I haven't ABX:ed it or anything,
but I just gives me sort of piece of mind, because I don't have to be worrying
about hearing some compression artefacts for making a choice to save some space
on the hard drive.

But still I draw the line somewhere about using overkill settings, so I don't use
-cbr 320

Yes this might sound crazy to some people, but to me it makes sense
weird me.

sorry I don't seem to know how to quote someone correctly, or did I do it right?

.veikko

(edit: added some text)

nah, it is overkill... --preset standard's goal is to be transparent and it is in most cases. If it isn't transparent it's because of a bug in LAME...so --preset extreme simply uses let's say 15kbps more for every frame than --preset standard would....just for heck of it... That sometimes prevents it from artificats...anyway, I don't feel like repeating myself...take a look here http://www.hydrogenaudio.org/forums/index....ndpost&p=276150 and here http://www.hydrogenaudio.org/forums/index....ndpost&p=276387

or use the search engine