Hydrogenaudio Forums > Why isn't recommended switch -q 0 in LAME?

yakkuru

Oct 17 2006, 12:42

Hi all,

do you know somebody why in LAME codec's help is recommended value 2 for -q switch (Noise shaping & psycho acoustic algorithms) while value 0 provides the highest quality? I have just made 2 mp3s: one with -q 2 and second with -q 0 setting but both was made in the same time. So I don't understand the reason why -q 2 is still recommended. Even in hydrogenaudio's FAQ there is no mention about this switch.

Thanks.

-------------------------------------------------------------------------------------------
Sorry my English and if this question is already answered. I couldn't fint it.

jmartis

Oct 17 2006, 12:56

Here I'm quoting my post from another thread:

QUOTE(jmartis @ Sep 21 2006, 18:23)

QUOTE(Irie @ Sep 21 2006, 18:08)

Alright, thanks!

By reading some more related topics I meanwhile learned that there was a bug regarding -q0 and -q1 in LAME versions 3.94-3.96.1. Besides the fact that they have been corrected in newer versions: what about -q 2, could there be any profits from using it in comparison to the default -q 3?

Sorry for probing questions, i'm just interested in these details

Irie

The artifact I mentioned exists at -q 0 up to -q 2 (it is described as "ringing") in ALL versions from 3.96, but it will appear just in extreme cases and I doubt you will hear any difference between -q 2 and -q 3. My recommendation is to use -q 3 (=default) quality setting.
(There can as well be cases where -q 2 (or -q 0) performs better than than default -q 3, but I have not noticed such case with my music)

Note: If you really want to know which is better for you, I suggest you to encode some of your favourite music and do an ABX test. This is the only way to find out.

J.M.

Junon

Oct 17 2006, 13:04

Deleted to prevent any further confusion. Author was obviously blind and therefore unable to distinguish between -q0 and -V 0. *sigh*

greynol

Oct 17 2006, 13:14

Junon,

I think you're confusing -q0 with -V 0.

Junon

Oct 17 2006, 13:23

QUOTE(greynol @ Oct 17 2006, 21:14)

Junon,

I think you're confusing -q0 with -V 0.

Damn! Yep, I really do, thanks for the error note. People, forget everything I wrote above! Seems like I'm a bit too silly to write helpful content this lousy day.

Firon

Oct 17 2006, 14:09

I recall hearing that -q0 has some sort of bug that causes artifacts, whereas -q2 doesn't. I don't know if said bug was ever fixed.

Mike Giacomelli

Oct 17 2006, 17:21

QUOTE(yakkuru @ Oct 17 2006, 11:42)

Hi all,

do you know somebody why in LAME codec's help is recommended value 2 for -q switch (Noise shaping & psycho acoustic algorithms) while value 0 provides the highest quality? I have just made 2 mp3s: one with -q 2 and second with -q 0 setting but both was made in the same time. So I don't understand the reason why -q 2 is still recommended. Even in hydrogenaudio's FAQ there is no mention about this switch.

Thanks.

-------------------------------------------------------------------------------------------
Sorry my English and if this question is already answered. I couldn't fint it.

As I recall, q0 is only very marginally better then q2, but much slower.

Also, in older versions of lame, q0 didn't work right, but that was fixed a while ago.

starcy

Oct 18 2006, 01:32

I am doing some ABX-ing and find that -q 0 and -m s work very good without and ringing

Zster

Oct 18 2006, 04:51

QUOTE(yakkuru @ Oct 17 2006, 20:42)

Hi all,

do you know somebody why in LAME codec's help is recommended value 2 for -q switch (Noise shaping & psycho acoustic algorithms) while value 0 provides the highest quality? I have just made 2 mp3s: one with -q 2 and second with -q 0 setting but both was made in the same time. So I don't understand the reason why -q 2 is still recommended. Even in hydrogenaudio's FAQ there is no mention about this switch.

Thanks.

-------------------------------------------------------------------------------------------
Sorry my English and if this question is already answered. I couldn't fint it.

This has in fact been answered numerous times by the developers. There is no difference between -q 0 and -q 2 when using the new VBR and settings lower than 2 only affects the old VBR mode in the current version of lame. If you don't believe me do a byte by byte comparison on the two files. After 3.96 -q0 broke for a bit but was fixed around 3.97 alpha 6. The value of q is automatically changed by changing the value of V when using the new vbr so there's no need to worry about it.

yakkuru

Oct 18 2006, 08:29

Thanks all for replies. Zster is right. At present LAME 3.97 there is no difference between -q 2 and -q 0 in new VBR mode. Thus -V x --vbr-new -h seems to be good option?

Raiden

Oct 18 2006, 10:17

QUOTE(yakkuru @ Oct 18 2006, 16:29)

Thanks all for replies. Zster is right. At present LAME 3.97 there is no difference between -q 2 and -q 0 in new VBR mode. Thus -V x --vbr-new -h seems to be good option?

According to the docs, -h is always enabled in VBR mode.

pdq

Oct 18 2006, 10:48

QUOTE(starcy @ Oct 18 2006, 03:32)

I am doing some ABX-ing and find that -q 0 and -m s work very good without and ringing

Then you should find -m j to be even better.

yakkuru

Oct 18 2006, 11:54

QUOTE(Raiden @ Oct 18 2006, 18:17)

According to the docs, -h is always enabled in VBR mode.

According to comparing method which Zster has advised (-q 0) - (-q 4) give same output but only in new VBR mode. In default/old VBR mode mp3 made with -h is different to mp3 made without -h argument.