Help - Search - Members - Calendar
Full Version: bug or feature ? --ape -V0 same as --ape -V1
Hydrogenaudio Forums > Lossy Audio Compression > MP3 > MP3 - General
bottom
hello

(sorry if this is going to b a newbie question,
but i havent found any similar postings
and just don't want to mess up my new mp3 songs)

since lame 3.90x i was using the following commandline:
--alt-preset extreme -V1 -Z

now with lame 3.95.1:
--preset extreme

but this resulted in extremely low bitrates on some songs
e.g. a classical solo guitar,
some rather complex orchestral tracks form debussy and vivaldi
and some jazz songs with massive 'high-hat cymbal' percussive elements
sometimes resulted in only about 180 kbs (!?)

(not hearing any problems,
but fearing quality loss) i tried (with 3.95.1):
--preset extreme -V1 (a) and
--preset extreme -V0 (b)

bitrates went up to about 190 but on SOME tracks i noticed that
tracks encoded with (a) and (b) had the exactly same bitrate
and their size differed only in 2 bytes !!??

(btw: for the tracks above lame 3.93.1 typical produces bitrates of 220-250)

- what am i doing wrong or has 3.95.1 been improved that much ?
or
- is there another 'sensible' way to raise bitrates for critical tracks ?
(lowering ath-level, -q0, ... ?)
and
- which presets do include the -Z option ?


thanks for your patience
(and any help)

bottom
gazzyk1ns
Adding things to the command line when using presets will only result in inferior quality, so it's obviously best to leave them as they are. It's a mistake a lot of us make when we first get into encoding so don't worry.

If you can't hear any problems then everything is fine - that's the simple answer. --alt-preset extreme (or --preset extreme... they're both the same) is highly tuned and is transparent (i.e. faultless) on most samples for most people. Even those with the best hearing can usually only hear artifacts under certain conditions, with certain samples. There are rare exceptions though. If you'd like to find out whether you really can hear a difference (which a lot of us here would recommend) then you should perform some ABX tests - see the sticky thread here for a guide.

If you really do need higher quality from your encodes, then you could try --alt-preset insane, which will give you a CBR 320 file and is the highest quality available from LAME. Be warned though, that if --ape doesn't encode your music transparently then --api probably won't either... if you're getting artifacts at this sort of bitrate then it's probably down to the limitations and flaws of the MP3 format. In that case, you should look at encoding to MPC or a lossless format such as FLAC.

Off the top of my head, I can't remember which of the presets use the -Z switch. When using the recommended version (3.90.3), though, you never need to worry about it - it's implemented 100% correctly. I'm sure someone else can tell you where it's used if you were just asking out of interest.
Pio2001
Z is not an option but a switch. I don't remeber what it does, but asking if it is used or not is the same thing as asking if someone changed the switch position of the light in the next room. Wether the answer is yes or no won't tell you if the light is on or off after the switch was switched, since you don't know if it was on or off before.
Z is too short a word to perform a search, so try finding posts about Lame 3.90.2 vs 3.90.3, since -Z was the main difference.
Gabriel
QUOTE
now with lame 3.95.1:
--preset extreme

but this resulted in extremely low bitrates on some songs
e.g. a classical solo guitar,
some rather complex orchestral tracks form debussy and vivaldi
and some jazz songs with massive 'high-hat cymbal' percussive elements
sometimes resulted in only about 180 kbs (!?)


What do you mean by extremely low bitrate? And by "sometimes"? Vbr modes are desiged to adjust bitrate according to content, so it is perfectly normal that sometimes bitrate is going down.

QUOTE
(not hearing any problems,
but fearing quality loss) i tried (with 3.95.1):
--preset extreme -V1 (a) and
--preset extreme -V0 (b)

bitrates went up to about 190 but on SOME tracks i noticed that
tracks encoded with (a) and (b) had the exactly same bitrate
and their size differed only in 2 bytes !!??


In fact your command lines (if you really used that) are equivalent to -V1 (a) and -V0 (b).
I am surprised to see an identical bitrate for both. Does your "SOME" really means "some" or "one"?
bottom
hi
and thanks for replying

[QUOTE]What do you mean by extremely low bitrate? And by "sometimes"? Vbr modes are desiged to adjust bitrate according to content, so it is perfectly normal that sometimes bitrate is going down.[/QUOTE]

'low' compared to 3.93.1 (~40 kbs difference)

'sometimes' means:
of course i experienced low bitrate with all previous lame version
e.g. modern piano music (no noisy background) hardly ever hit 200 kbs with --ape
but 3.95.1 even compressed music with (imho) complex structures in it (orchestra, percussion, ...) to a comparable low level

[QUOTE]In fact your command lines (if you really used that) are equivalent to -V1 (a) and -V0 (b).
I am surprised to see an identical bitrate for both. Does your "SOME" really means "some" or "one"? QUOTE]

using 3.95.1 (is there are short form for this ?) only since last week and i havent double encoded every song two times (got a pII 233 running here, batch converting over night). but i found the following songs with nearly identical size, absolutely identical size or size differing by only 1 or 2 bytes:
- Antonio Vivaldi - Concerto in f-minor (I Musici).wav
(3 short tracks, total runtime 5min , baroque orchestra) (average) bitrate 200
- A. Barrios - Una Limosna por el amor de Dios.wav
(4min , guitar) bitrate 220
- Claude Debussy - 'Syrinx'.wav
(3min, flute) bitrate 180
- Billy Cobham - Culture Mix (Intro).wav
(4min, percussive introduction, live recording) bitrate ~180

while i can perfectly live with this low bitrates (if quality is ok !?)
i found/find it rather strange that -V0 and -V1 doesn't really produce significantly different bitrates (like all previous lame versions i tested had done (3.87, 3.88betas, 3.90, 3.91, 3.93)

btw:
i'm using the 3.95.1 mitiok compile: http://mitiok.cjb.net/
(it still says "3.95" when encoding, but since the help files and file dates are updated i suppose it has to be the latest version)

btw1:
when saying "not hearing any problems" i mean even if i would have a good hearing, there is no way to actual make use of it here in front of my pc:
loud environment, rather 'cheap' soundcard, extremely bad speakers, headphones dead, ...
hope i can effort a better system in the near future... (sigh)

btw2:
> lame --longhelp (3.95.1 !!!)
experimental switches:
...
-Z [n] toggles the scalefac-scale and subblock gain feature on
if n is set and minus, only scalefac-scale is enabled

btw3:
[QUOTE]In fact your command lines (if you really used that) are equivalent to -V1 (a) and -V0 (b).[/QUOTE]
do you mean:
"--preset extreme -V0" is the same as "-V0" and
"--preset extreme -V1" is the same as "-V1" ???


bottom
Gabriel
QUOTE
do you mean:
"--preset extreme -V0" is the same as "-V0" and
"--preset extreme -V1" is the same as "-V1" ???

In fact it should, but I just tested and it is not the case actually. I will have to investigate this.
phwip
QUOTE(Gabriel @ Jan 25 2004, 01:51 PM)
QUOTE
do you mean:
"--preset extreme -V0" is the same as "-V0" and
"--preset extreme -V1" is the same as "-V1" ???

In fact it should, but I just tested and it is not the case actually. I will have to investigate this.

The reason that these command lines should be equivalent is because in 3.95.1, --preset extreme is just a synonym for -V0. Encode a file with either and you get the same output. In fact all the vbr presets map directly onto -V switches.

So, --preset extreme -V0 should be just another way of writing -V0 -V0
and --preset extreme -V1 should be the same as -V0 -V1

In the case of using multiple -V values in a command line, one is redundant. So the first is discarded. -V0 -V0 is just -V0. And -V0 -V1 is the same as -V1.

The problem that Gabriel is having to check is that for some reason --preset extreme -V1 is not actually giving the same output as just -V1.
Gabriel
QUOTE
The problem that Gabriel is having to check is that for some reason --preset extreme -V1 is not actually giving the same output as just -V1.


I found the source, and this behaviour is by design. Presets are enforcing parameters, while default settings are not. So what is happening is that --preset extreme is positionning some value, and when it comes to positionning default values from -V1, as some are parameters are already set, they are not modified.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.