How does --alt-preset standard compare to my standard command line:
-b 160 -m s -h -p -V 0 -B 160 -F -t --vbr-new -Y
-b 256 -m s -h -p -V 0 -B 256 -F -t --vbr-new -Y
I have been perusing the site and it seems as if --alt-preset standard (aps) is the recommended LAME setting. Does aps use the new VBR encoding? Does apr include CRC-checking? And what is the -Y switch used for (I included the switch based on forum examples)? Does aps incorporate --nspsytune? Should it be b160 or b 160 (inclusion of a space)?
And what of the following option strings:
--alt-preset standard -p -Y -V 0
--alt-preset extreme -p -Y -V 0
I was thinking that the extreme example would serve my purposes as an all-around setting...
With those bitrate limits it looks like CBR...
Use --alt-preset standard, extreme or medium.
Plenty of arguments on this forum already
I have been perusing the site and it seems as if --alt-preset standard (aps) is the recommended LAME setting.
eagle-eye strikes again welcome to the forum.
Does aps use the new VBR encoding?
no. which is not a bad thing. vbr new is faster, but not necessary of higher quality.
Does apr (...)
apr? typo?
How does --alt-preset standard compare to my standard command line:
-b 160 -m s -h -p -V 0 -B 160 -F -t --vbr-new -Y
-b 256 -m s -h -p -V 0 -B 256 -F -t --vbr-new -Y
your commandlines do not make that much sense to me, as you use e.g. '-b 160' and '-B 160', what really is kind of useless AFAIK. also '-m s' instead of joint stereo.
the presets have been tuned to give you the best possible quality in general. USE THEM.aps is of higher quality than your custom lines.
btw, it is not considered nice to edit posts other than for wrong spelling and wrong informations. make a new post next time in order to get a working discussion.
Will do...And thanks for the welcome =>)! Just a few more questions...
1. Should it be b160 or b 160 (inclusion of a space)?
2. Can options be strung together as in -xyz?
I was thinking that the following examples would serve my purposes as all-around settings:
--alt-preset standard -p -V0 -Y -t
--alt-preset extreme -p -V0 -Y -t
1. In the case of aps, is V being set to 0? What about ape?
2. Are VBR tags being written to the file in aps? In ape?
3. Does -Y reduce file size at the expense of frequencies over 16kHz? If so, then I would not use the switch in aps, but only in ape...
I am wondering why your are using switches without knowing their behaviour or signification.
You would better use the unaltered presets.
Regarding your first question, the answer is clear: alt-preset standard is better than your custom command lines.
I was thinking that the following examples would serve my purposes as all-around settings(...)
the presets have been tuned to give you the best possible quality in general.
if you did not notice any artifacts, use the presets without modifications regarding sound quality (there are known exeptions, browse the forum if intersted).
1. In the case of aps, is V being set to 0? What about ape?
huh? do have a real idea of what you are talking about here?
2. Are VBR tags being written to the file in aps? In ape?
aps=vbr. ape=also vbr. conclusion?
3. Does -Y reduce file size at the expense of frequencies over 16kHz? If so, then I would not use the switch in aps, but only in ape...
and that is why?
1. In the case of aps, is V being set to 0? What about ape?
huh? do have a real idea of what you are talking about here?
Oops...Answered here: <http://www.hydrogenaudio.org/forums/index.php?showtopic=18091>. aps is to V2, ape to V0 by default.
2. Are VBR tags being written to the file in aps? In ape?
aps=vbr. ape=also vbr. conclusion?
I was wondering if the LAME tag was written to the file by default. I enabled this switch to disable writing the tag (thereby reducing filesize/overhead; the filesize/overhead savings possibly being an inconsequential issue).
3. Does -Y reduce file size at the expense of frequencies over 16kHz? If so, then I would not use the switch in aps, but only in ape...
and that is why?
Heh...Wanted to know if -Y reduces filesize or simply allow bits to be used elsewhere. That way I can either apply the switch to the reduction of bits within a certain frequency range or to further reduce filesizes. Excuse the confusion !
He's put "-p" in there as well.. I know this has been posted elsewhere (but I cant blinkin' find it!).. what does that switch do again?
He's put "-p" in there as well.. I know this has been posted elsewhere (but I cant blinkin' find it!).. what does that switch do again?
[a href="index.php?act=findpost&pid=256895"][{POST_SNAPBACK}][/a]
Enables CRC "error protection. [A]dds 16 bit checks (the checksum is computed correctly)" to help verify files during transfers (i.e file sharing).
I enabled this switch to disable writing the tag (thereby reducing filesize/overhead; the filesize/overhead savings possibly being an inconsequential issue).
You wan to remove the Info tag to save space but at the same time you are enabling crc check (-p)??? That is quite inconsistant.
Enables CRC "error protection. [A]dds 16 bit checks (the checksum is computed correctly)" to help verify files during transfers (i.e file sharing).
The mp3 frames crc will not change anything regarding file sharing, unless the file sharing application is checking this specific crc. But in practice, the do not check this crc.
This only informs a decoder that a frame is corrupted, so it can ask retransmission if possible, or use interpolation/muting/repetition strategies.
Once again, if you do not understand the meaning of an option, do not use it.
The thing is, the presets that are suggested on this site aren't anywhere near the best quality you can get from lame. No where near! Absolutely no work has been done on them, no one has tested them - they were just thrown together on a quiet Sunday afternoon, posted up here, and no one has ever checked them since.
Read through the list of lame command options, throw a few random ones together, use that setting and it will be far better than the "standard" and "extreme" presets suggested here. Your settings will actually work even better if you don't have a clue what the switches do - just try random letters and numbers and you'll get the best mp3s ever!
Whatever you do, don't use the search function, and don't read the sections in the FAQ which try to suggest that the presets are the best you can possibly do. All that stuff, explaining why "adding extra switches to the presets will almost certainly reduce quality" - ignore it - it's all lies!
Cheers,
David.
P.S. If anyone needs to sign up for the course "how to spot sarcasm a mile off" please PM me.
Another entertaining thread, but I seriously wonder how incidents like this happen? How come some people that haven't got a clue, think they can outsmart everyone, even the developers? They hopefully know they couldn't develop an encoder themselves, and maybe they don't even know the first thing about the technologies involved, but they still think they know more about how to use the "finished" product, than the people that made it.
What is it that makes them think that common sense doesn't apply at all? How likely is it that an hours random tuning done by a random person, will seriously improve something that a group of people have been working hard on for years?
I bet a scientist would love to get hold of a specimen for experiments. It's likely that finding the mechanism that triggers such out-of-control self confidence, would be pretty useful for military purposes.
The sad part is that they have the interest, but waste their time doing useless "tuning". Time that would be alot better spent reading the FAQ, old posts, links and so on. And if they don't care about learning, they still make files with inferior quality.
Edit: spelling
Well, now adamlau will be afraid of ever posting here again?
Hopefully. At least before he does a lot of reading.
The thing is, the presets that are suggested on this site aren't anywhere near the best quality you can get from lame. No where near! Absolutely no work has been done on them, no one has tested them - they were just thrown together on a quiet Sunday afternoon, posted up here, and no one has ever checked them since.
Read through the list of lame command options, throw a few random ones together, use that setting and it will be far better than the "standard" and "extreme" presets suggested here. Your settings will actually work even better if you don't have a clue what the switches do - just try random letters and numbers and you'll get the best mp3s ever!
Whatever you do, don't use the search function, and don't read the sections in the FAQ which try to suggest that the presets are the best you can possibly do. All that stuff, explaining why "adding extra switches to the presets will almost certainly reduce quality" - ignore it - it's all lies!
Cheers,
David.
P.S. If anyone needs to sign up for the course "how to spot sarcasm a mile off" please PM me.
[a href="index.php?act=findpost&pid=256903"][{POST_SNAPBACK}][/a]
PS: Can't help myself...
Edit: 2Bdecided's article has to be added to the FAQ, what do you think of it Pio2001?
On the Mac side at least, with special regard to iTunes-LAME, many people are misled because of the documentation that comes with the LAME CLI. For example, a number of people have posted on the iTunes LAME forum along the lines of 'I use [insert horrible flags here] commandline which sounds great to me' and 'presets are bad because they don't give you full control', or that -q 0 + -V 0 == best, and stuff like that. A lot of misinformation also shows up when people are switching from other encoding engines, ie. ms vs. js, etc.
Yeah, it seems like when it comes to computers, a lot of people like to figure things out for themselves. Thing is, that is totally inconsistant with using a forum, in which other people have already figured something out and there's community consensus on it. Its always boggling when the people that use these home grown command lines even bother posting about it--its just backwards.
EDIT:
adamlau: Sorry buddy. You should just use the presets (or V settings) with nothing else added.
No harm, no foul in the quest for clarification...And yes I have read through --longhelp and the FAQ and more than a few forum posts...I posted because I still did not (and still do not) understand...
--alt-preset extreme -Y -t
1. Does -Y reduces filesize at the expense of a certain frequency range or simply allow bits to be used elsewhere (determines whether I retain -Y)?
2. Is the LAME tag was written to the file by default (determines whether I should leave -t)?
BTW, I will never be afraid to ask questions ...Thanks all!
2. Is the LAME tag was written to the file by default (determines whether I should leave -t)?
[a href="index.php?act=findpost&pid=257060"][{POST_SNAPBACK}][/a]
Yes, the LAME tag is written by default. But using -t to disable it is useless for VBR, because it will also disable the VBR information needed for correct time and seeking.
Excellent...Now about the -Y switch ...
http://www.hydrogenaudio.org/forums/index....7516#entry74068 (http://www.hydrogenaudio.org/forums/index.php?act=ST&f=5&t=7516#entry74068)
Rather than going through and answering your questions I will leave it up to the search function where you will get full & indepth answers.
Sometimes I wonder if someone isn't playing a twisted trick on us...
Oh thanks 2BDecided and others. It made the thread worth ...well... I dunno ..better?
Gabriel, you have my respect.
<snip>
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=256903")
Oh, by all means, use the LAME documentation to find the settings right for you. Because hey, when the command line help, the [a href="http://lame.sourceforge.net/USAGE]USAGE[/url], the manual (http://lame.sourceforge.net/doc/html/index.html), and the examples (http://lame.sourceforge.net/doc/html/examples.html) say
exactly the same thing, who are you going to trust - the developers who wrote the documentation, or a bunch of forum hecklers?
Is the LAME tag was written to the file by default (determines whether I should leave -t)?
Yes, it is written by default. But if you disable it, remember that you are also loosing the ReplayGain info and the gapless info.
Oh, by all means, use the LAME documentation to find the settings right for you. Because hey, when the command line help, the USAGE, the manual, and the examples say exactly the same thing, who are you going to trust - the developers who wrote the documentation, or a bunch of forum hecklers?
Well, the USAGE file would need to be updated.
But it seems to me that the html documentation is not promoting "exotic" switches, is it?
Well, the USAGE file would need to be updated.
But it seems to me that the html documentation is not promoting "exotic" switches, is it?
[a href="index.php?act=findpost&pid=257111"][{POST_SNAPBACK}][/a]
Well, the examples I linked above are perpetrating one pure myth (-h) and something I'm not sure is particularly useful to anybody anymore and might be a myth too (-f). The HTML command line reference, and all the command line references in my opinion, are going over options that really have no point whatsoever and are just confusing people right now, because their descriptions are hopelessly incorrect. Examples: -h, --allshort, --athtype, -m, --short, --vbr-old, --vbr-new, -X. All these things are either completely obviated by other things or basically shouldn't be used by anybody. So why have them? Plus most of the rest of the options, while quite useful to some people, are easily abused.
More generally, I see this documentation gap where it isn't quite drilled deep enough into peoples' heads that they
need to be using the presets unless they have good objective reasons not to. Presets aren't nearly as prominent as they should be. If somebody comes out of reading the manual and
ever gets the idea that it's better to roll his/her own command line for transparent encoding rather than using one of the presets, without ABXing, I'd say that is fundamentally a documentation problem.
This is why there is a basic switch reference and a full switch reference.
The basic options should be safe.
The full reference is there because otherwise people complain because some switches are not documented. I should perhaps add a disclaimer to the full list.
This is why there is a basic switch reference and a full switch reference.
The basic options should be safe.
The full reference is there because otherwise people complain because some switches are not documented. I should perhaps add a disclaimer to the full list.
[a href="index.php?act=findpost&pid=257127"][{POST_SNAPBACK}][/a]
Yes, please make clear which switches are safe to use, which are experimental and/or which have known issues.
<snip>
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=256903")
Oh, by all means, use the LAME documentation to find the settings right for you. Because hey, when the command line help, the [a href="http://lame.sourceforge.net/USAGE]USAGE[/url], the manual (http://lame.sourceforge.net/doc/html/index.html), and the examples (http://lame.sourceforge.net/doc/html/examples.html) say exactly the same thing, who are you going to trust - the developers who wrote the documentation, or a bunch of forum hecklers?
[a href="index.php?act=findpost&pid=257075"][{POST_SNAPBACK}][/a]
I've got lost now - remind me - which of us was being sarcastic?
Cheers,
David.
I've got lost now - remind me - which of us was being sarcastic?
Cheers,
David.
[a href="index.php?act=findpost&pid=257171"][{POST_SNAPBACK}][/a]
Both of us?
This is why there is a basic switch reference and a full switch reference.
The basic options should be safe.
The full reference is there because otherwise people complain because some switches are not documented. I should perhaps add a disclaimer to the full list.
[a href="index.php?act=findpost&pid=257127"][{POST_SNAPBACK}][/a]
Yes, I think a disclaimer and a dichotomy of options is definitely the way to go. I'm mostly bickering about the style of how it is done, rather than the overall organization. Specifically, the basic command line reference doesn't mention the presets at all also mentions options that should never be used (-h and -f). The long command line reference should only be something that people refer to when they have specific hardware or software requirements, not when trying to tweak quality settings.
Specifically, the basic command line reference doesn't mention the presets at all
[a href="index.php?act=findpost&pid=257180"][{POST_SNAPBACK}][/a]
Maybe this is because from Lame 3.95 on you do not need the --presets any more as all CBR, ABR and VBR settings are mapped to the presets. Actually, if you want to use something in between --preset medium and --preset standard you can only access this mode via -V3. I think this preset mapping was one of the best decisions since the "invention" of Lame, as probably most users dealing with Lame 3.90-3.93 don't know about the presets and use the standard gpsycho mode, mostly with "True" (haha) stereo. I think the best way would be to drop the talk about "presets" in the documentation since they are used by default anyway. It should be pointed out that the best quality is achieved with the VBR modes, with -V2 being the "recommended" setting if transparency is desired. Also the target bitrates of the V modes should be mentioned, IMHO. After this, ABR and CBR modes should be briefly mentioned and their disadvantages pointed out. It should also be explicitly stated that no extra options like -m s or practically anything should be used because it would lead to inferior results.
mentions options that should never be used (-h and -f)
Why should they never be used? Sometimes you need more speed.
I guess I was mistaken about the presets - -V is definitely the better way to go. I was under the impression that neither -h nor -f were recommended nor actively developed or tested, I could definitely be mistaken.
CRC checks are more or less completely useless unless you use them for verifying during streaming... and since they take up space they actually lower quality.
CRC checks are more or less completely useless unless you use them for verifying during streaming... and since they take up space they actually lower quality.
[a href="index.php?act=findpost&pid=257265"][{POST_SNAPBACK}][/a]
I was about to call BS on that one, but decided to check to make sure. Sure enough, the increase in size does not match exactly the extra 16-bit CRC per frame. Not only that, but the bit reservoirs are changed. The files are clearly not encoded the same.
Now that blows my mind, because I figured the file would just be encoded as normal with checksums computed and just stuck on after a frame is encoded, without affecting the bit reservoir. Interesting indeed. Now that I think about it I suppose this would make sense for CBR and ABR, since the encoder is more or less confined to a certain bitrate. I figured VBR wouldn't care, but apparently it works the same--the CRC affects all modes.
So... What exactly is wrong with -h? -q0 I know about, but -h is equivalent to -q2, and I've never heard anything bad about that. Heck, I even posted a while back troubleshooting a commandline with -h in it and Gabriel himself didn't say a thing about it.
In 3.96.1, the VBR presets default to q3, so you'd be changing that...
For the presets, sure, it makes sense. They chose q3 because nobody could ABX between Q2, and it saved time. But let's say the slightly lesser presets. I encode MP3s of my band for internet download using the -V5 --athaa-sensitivity 1 setting, because it saves bandwidth and the time of people who are downloading it. What harm, if any, would using the -h switch in addition to the above do?
Different people will probably have some different opinions, but I personnaly see nothing wrong in using -h with 3.96.1.
If the presets all use q3 but -h (q2) is recommended in the documentation for high quality encoding. It makes me feel like I should always use the -Vx switch in combination with -h.
The confusion needs clearing up.
If the presets all use q3 but -h (q2) is recommended in the documentation for high quality encoding. It makes me feel like I should always use the -Vx switch in combination with -h.
The confusion needs clearing up.
[a href="index.php?act=findpost&pid=257321"][{POST_SNAPBACK}][/a]
Unfortunately, me too!
Interesting... Maybe I'll do some ABX tests when I get home. I am dreading it though; ABXing at this level is extremely fatiguing...
They chose q3 because nobody could ABX between Q2, and it saved time.
[a href="index.php?act=findpost&pid=257309"][{POST_SNAPBACK}][/a]
In recent LAME versions
-q3 is the equivalent of the old
-q2 (aka
-h). v
3.94b introduced a new
-q option and that bumped the old options' numbers by one. I.e.
-q0 became
-q1,
-q1 became
-q2 and so on.
LAME 3.94 beta December 15 2003
* Takehiro Tominaga:
o fixed block switching of nspsytune
o best huffman divide in the inner loop. This should improve the quality,
but PAINFULLY slow. So it is not enabled by default. Use -q0 to use it.
o Changed -q option mapping. "-q2" until version 3.93 is now "-q3".
They chose q3 because nobody could ABX between Q2, and it saved time.
[a href="index.php?act=findpost&pid=257309"][{POST_SNAPBACK}][/a]
In recent LAME versions -q3 is the equivalent of the old -q2 (aka -h). v3.94b introduced a new -q option and that bumped the old options' numbers by one. I.e. -q0 became -q1, -q1 became -q2 and so on.
LAME 3.94 beta December 15 2003
* Takehiro Tominaga:
o fixed block switching of nspsytune
o best huffman divide in the inner loop. This should improve the quality,
but PAINFULLY slow. So it is not enabled by default. Use -q0 to use it.
o Changed -q option mapping. "-q2" until version 3.93 is now "-q3".
[a href="index.php?act=findpost&pid=257888"][{POST_SNAPBACK}][/a]
I've been reading here that the new qval 0 is not working well in newer verions of
lame. 3.95.x on newer. Please can some one clear this up. I though that q2 gave
the best speed:quality. Q1 is even higher quality but slower and q0 is the very
best but also very, very slow.
Is this q3 what q2 was and q0 is now even higher quality ?
Thank in advance and this is the best most informitive site ever. Is it also possible
to note the qval settings in the FAQ ?
Okay, let's break 'em down...
-b 160 -m s -h -p -V 0 -B 160 -F -t --vbr-new -Y
Just plain stupid. by setting min and max bitrate to 160kbps, you've essentially told Lame to make a 160kbps CBR MP3. The -p is harmful to quality and probably useless for your purposes.
-b 256 -m s -h -p -V 0 -B 256 -F -t --vbr-new -Y
This is dumb for all the same reasons except this time you will essentially get a 256kbps CBR MP3.
--alt-preset standard -p -Y -V 0
Why are you so intent on using -p and -V 0? -V 0 will mostly just make the MP3's bigger (if you actually let them be VBR unlike in your previous example).
--alt-preset extreme -p -Y -V 0
[a href="index.php?act=findpost&pid=256877"][{POST_SNAPBACK}][/a]
what about this one;
--abr 256 -q0 --nspsytune --athtype 2 --lowpass 20.0 --ns-bass -6
I've seen a few "experts" who recommend it for optimal results at 256kbps, yet I have a feeling it's not..
what about this one;
--abr 256 -q0 --nspsytune --athtype 2 --lowpass 20.0 --ns-bass -6
[a href="index.php?act=findpost&pid=261869"][{POST_SNAPBACK}][/a]
this is just ridiculous! If someone really likes to have such high bitrates, it would be the best to just use --preset extreme
Repeat after me: the --alt-presets are there for a reason! These other switches DO NOT work better then it, trust me on this.
what about this one;
--abr 256 -q0 --nspsytune --athtype 2 --lowpass 20.0 --ns-bass -6
With 3.96.1, if you want to achieve 256kbps abr, you should use:
--abr 256