Skip to main content
Topic: LAME versions (Read 33995 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

LAME versions

Reply #25
Quote
But like to know the huffman coding is making the files small. (or have i got this wrong   )
[a href="index.php?act=findpost&pid=258117"][{POST_SNAPBACK}][/a]


Not sure if I got you right here... The Huffman coding kicks at the end of the encoding process after most of the "squeezing" took place.

Quote
The Huffman Coding

At the end of the perceptual coding process, a second compression process is run. However, this second round is not a perceptual coding, but rather a more traditional compression of all the bits in the file, taken together as a whole. To use a loose analogy, you might think of this second run, called the " Huffman coding," as being similar to zip or other standard compression mechanisms (in other words, the Huffman run is completely lossless, unlike the perceptual coding techniques). Huffman coding is extremely fast, as it utilizes a look-up table for spotting possible bit substitutions. In other words, it doesn't have to "figure anything out" in order to do its job.

The chief benefit of the Huffman compression run is that it compensates for those areas where the perceptual masking is less efficient. For example, a passage of music that contains many sounds happening at once (i.e., a " polyphonous" passage) will benefit greatly from the masking filter. However, a musical phrase consisting only of a single, sustained note will not. However, this passage can be compressed very efficiently with more traditional means, due to its high level of redundancy. On average, an additional 20% of the total file size can be shaved during the Huffman coding.


Quote
The Huffman coding:

The MP3 also uses the classic technique of the Huffman algorithm. It acts at the end of the compression to code information, and this is not therefore itself a compression algorithm but rather a coding method.

This coding creates variable length codes on a whole number of bits. Higher probability symbols have shorter codes. Huffman codes have the property to have a unique prefix, they can therefore be decoded correctly in spite of their variable length. The decoding step is very fast (via a correspondence table). This kind of coding allows to save on the average a bit less than 20% of space.

It is an ideal complement of the perceptual coding: During big polyphonies, the perceptual coding is very efficient because many sounds are masked or lessened, but little information is identical, so the Huffmann algorithm is very seldom efficient. During "pure" sounds there are few masking effects, but Huffman is then very efficient because digitalized sound contains many repetitive bytes, that will then be replaced by shorter codes.

LAME versions

Reply #26
Ah cool thanks, Does Lame do huffman coding with the presets ?
Death is the one thing we all face

LAME versions

Reply #27
Quote
I think it would be realy cool if lame could take a second pass if say --mp3gain
was enabled to change the files gain to avoid the clipping.

Also is the q 0 switch fixed. I have no real regard for encoding time. But like
to know the huffman coding is making the files small. (or have i got this wrong   )
[a href="index.php?act=findpost&pid=258117"][{POST_SNAPBACK}][/a]

That's a feature I'm dreaming about all the time, because now I can only do MP3Gain modification which is lowering the volume too much from the original.
Or I can use --scale 0.9X but I could never know if the result will be without clipping. That's too much pain to achieve my goal

I the other hand that's true that HW players ignore ReplayGain informations, the only thing how to avoid clipping is MP3Gain modification AFAIK...

Yes, the fixed -q 0 switch would be nice...
I still don't have a clue why there are the options of -q 0 and -q 1, when they are not safe and flawless...
Sorry for my poor English, I'm trying to get better... ;)
"The greatest trick the Devil ever pulled, was convincing the world he didn't exist."

LAME versions

Reply #28
Quote
That's a feature I'm dreaming about all the time, because now I can only do MP3Gain modification which is lowering the volume too much from the original.
Or I can use --scale 0.9X but I could never know if the result will be without clipping. That's too much pain to achieve my goal

I the other hand that's true that HW players ignore ReplayGain informations, the only thing how to avoid clipping is MP3Gain modification AFAIK...

Yes, the fixed -q 0 switch would be nice...
I still don't have a clue why there are the options of -q 0 and -q 1, when they are not safe and flawless...


I see i never thought of using --scale. Thanks. Does scale time each samples
value by 0.n making less chance of a sample over 32768 ?
Death is the one thing we all face

LAME versions

Reply #29
Quote
I see i never thought of using --scale. Thanks. Does scale time each samples
value by 0.n making less chance of a sample over 32768 ?
[a href="index.php?act=findpost&pid=258150"][{POST_SNAPBACK}][/a]

Yes, I think it's quite obvious, but you can never guess if it is enough to avoid clipping or if it isn't too much...

Edit: So you have to encode with some --scale setting and then verify it by MP3Gain if it is clipping free... And so on...
Sorry for my poor English, I'm trying to get better... ;)
"The greatest trick the Devil ever pulled, was convincing the world he didn't exist."

LAME versions

Reply #30
Quote
Quote
I see i never thought of using --scale. Thanks. Does scale time each samples
value by 0.n making less chance of a sample over 32768 ?
[a href="index.php?act=findpost&pid=258150"][{POST_SNAPBACK}][/a]

Yes, I think it's quite obvious, but you can never guess if it is enough to avoid clipping or if it isn't too much...

Edit: So you have to encode with some --scale setting and then verify it by MP3Gain if it is clipping free... And so on...
[a href="index.php?act=findpost&pid=258151"][{POST_SNAPBACK}][/a]


I will test it out this week end. I do not mind the odd sample. But encoding
"All becuase of you" from U2 gives terrible clipping at 98db. I do not undestand
why newer cd are masterd so loud.

Just like to say great work with lame. I can not wait to test the beta of 4.0
when it hapens
Death is the one thing we all face

LAME versions

Reply #31
Quote
Yes, the fixed -q 0 switch would be nice...
I still don't have a clue why there are the options of -q 0 and -q 1, when they are not safe and flawless...

They are fixed in current alpha version.

LAME versions

Reply #32
Quote
Well, if someone wants to take over the frontends for v4, the position is open.
[a href="index.php?act=findpost&pid=258077"][{POST_SNAPBACK}][/a]

great! So LAME will finally come with a GUI!

Quote
Quote
Sorry, if I missed this, but is Intensity Stereo planned for LAME 4.0?

Yes
[a href="index.php?act=findpost&pid=258105"][{POST_SNAPBACK}][/a]

I just learned that Intensity Stereo is lossy...what would be the use for that? Doesn't Joint-Stereo work good enough already?
--alt-presets are there for a reason! These other switches DO NOT work better than it, trust me on this.
LAME + Joint Stereo doesn't destroy 'Stereo'

LAME versions

Reply #33
Quote
Quote
Well, if someone wants to take over the frontends for v4, the position is open.
[a href="index.php?act=findpost&pid=258077"][{POST_SNAPBACK}][/a]

great! So LAME will finally come with a GUI!
[a href="index.php?act=findpost&pid=258160"][{POST_SNAPBACK}][/a]

frontend != GUI
Quote
Quote
what you mean by frontend?

The frontends are lame.exe, lame.dll, acm codec, directshow filter.
[a href="index.php?act=findpost&pid=258116"][{POST_SNAPBACK}][/a]

LAME versions

Reply #34
Quote
great! So LAME will finally come with a GUI!

No

LAME versions

Reply #35
Well...Lame 4.0 alpha 11 seems to be very good with APS... 

Gabriel,can u tell me why qval is locked with vbr?

LAME versions

Reply #36
Quote
I just learned that Intensity Stereo is lossy...what would be the use for that? Doesn't Joint-Stereo work good enough already?

Intensity Stereo should be better than Mid-Side Stereo for low bitrates.

By Joint Stereo I assume you mean Lames Mid-Side stereo implementation. Both Mid-Side Stereo and Intensity Stereo are Joint Stereo.

LAME versions

Reply #37
Quote
You're all also forgetting Takehiro's work on LAME4.0. This is currently at Alpha 11.

Last time I did this, I got into trouble  , but here goes -  you can d/l an exe of the latest 4.0 alpha 11 build from: http://homepage.ntlworld.com/jfe1205/lame4.0a11.zip. But please remember this is an alpha build for testing only. This is purely for people to see how things are going. This is not to be used for archival, nor do I believe Takehiro is particulalry interested in detailed feedback at this time. When he's ready for this, he'll request it.
[a href="index.php?act=findpost&pid=256157"][{POST_SNAPBACK}][/a]


Wow that Lame4 alpha is a blindingly fast encoder. About 4 times as fast as 3.96.1 and nearly 8 times as fast as 3.90.3 on my system.  I'll be watching for that to become beta or release.

LAME versions

Reply #38
Quote
Quote
You're all also forgetting Takehiro's work on LAME4.0. This is currently at Alpha 11.

Last time I did this, I got into trouble  , but here goes -  you can d/l an exe of the latest 4.0 alpha 11 build from: http://homepage.ntlworld.com/jfe1205/lame4.0a11.zip. But please remember this is an alpha build for testing only. This is purely for people to see how things are going. This is not to be used for archival, nor do I believe Takehiro is particulalry interested in detailed feedback at this time. When he's ready for this, he'll request it.
[a href="index.php?act=findpost&pid=256157"][{POST_SNAPBACK}][/a]


Wow that Lame4 alpha is a blindingly fast encoder. About 4 times as fast as 3.96.1 and nearly 8 times as fast as 3.90.3 on my system.  I'll be watching for that to become beta or release.
[a href="index.php?act=findpost&pid=263128"][{POST_SNAPBACK}][/a]

Holy Cow!
It took only 17 sec. to encode a 4 minute song on my P4-3,4Ghz.
(with --preset cbr 128) Well, I know he doesn't want any feedback but this is so amazing!
portable: 128 kbps cbr AAC
local: -7 FLAC

LAME versions

Reply #39
Quote
Holy Cow!
It took only 17 sec. to encode a 4 minute song on my P4-3,4Ghz.


Maybe it's more highly optimised for the Athlon at the moment becuase I'm getting nearly double that speed (about 8 seconds to encode a 4 minute track) on my AthlonXP 2800+. That was using vbr (-V4) which is the same as the old --alt preset medium. The default cbr 128 is much the same speed.

Hang on, I just found the "-h" (high quality mode) option, yeah that slows down the cbr encoding quite a bit. My cbr encoding time goes up to about 16 seconds (4 min track) with that option. It doesn't seem to have any effect on my vbr ecoding times however, which are still amazingly fast at about 7 or 8 seconds per 4 minute track !

LAME versions

Reply #40
just wanted to give out a short "thank You" to Takehiro, Gabriel and the other LAME people that probably don't post here

by the way, what ever happened to Naokis nspytune2 before he retired?
I just read something about it on his old dev notes.

LAME versions

Reply #41
It encoded a 7 minute song in about 20 seconds using -V 2...!

I tested fatboy at 128 kbps using 4.0 alpha 11 and 3.96.1, and the latter definately has better quality, though of course neither is transparent. Certainly, it's only an alpha, so what can you expect?

Not transparency, that's what.

LAME versions

Reply #42
I found it pretty interesting that LAME 4.0 alpha --preset standard lowpasses at ~17000 hz but still produces a higher bitrate than LAME 3.96.1 even though it lowpasses at ~ 19000 hz...I always thought high frequencies are the 'most expensive' ones...so that doesn't make any sense to me...
--alt-presets are there for a reason! These other switches DO NOT work better than it, trust me on this.
LAME + Joint Stereo doesn't destroy 'Stereo'

LAME versions

Reply #43
4.0 is not yet tuned.
You can play with it, but I would advise you to NOT USE IT for production.

 

LAME versions

Reply #44
Quote
4.0 is not yet tuned.
You can play with it, but I would advise you to NOT USE IT for production.
[a href="index.php?act=findpost&pid=263395"][{POST_SNAPBACK}][/a]


Gabriel, why is there development of LAME v4.0 whilst LAME 3.97 is still being developed?  From a dev lifecycle point of view it doesn't seem logical to go from 3.97 to 4.0 (I know neither are released yet or indeed in beta form).  Does 4.0 have everything that 3.97 has?... but more.  Will ther be 3.98, 3.99?

I guess what I'm really trying to get at is, why invest time to work on and release 3.97 when those developers could work on 4.0 and possibly get that version out sooner? 

BTW this isn't me having a go or anything, I'm simply interested from a LAME progression point of view, and as a developer myself.

Thanks

jb

LAME versions

Reply #45
Lame 3.xx is reaching the end of its possible enhancements, regarding both speed and quality.
The current goal of v4 is to build a new solid basis removing current limitations as much as possible. Right now the v4 task is more an architectural one than a tuning or polishing one.
Thus it is not very important if v4 doesn't have the same tunings as the latest 3.xx versions.

LAME versions

Reply #46
Quote
Lame 3.xx is reaching the end of its possible enhancements, regarding both speed and quality.
The current goal of v4 is to build a new solid basis removing current limitations as much as possible. Right now the v4 task is more an architectural one than a tuning or polishing one.
Thus it is not very important if v4 doesn't have the same tunings as the latest 3.xx versions.
[a href="index.php?act=findpost&pid=263458"][{POST_SNAPBACK}][/a]


Great, thanks.  Makes sense.

LAME versions

Reply #47
3.97alpha5 is now available at Rarewares.
John
----------------------------------------------------------------
My compiles and utilities are at http://www.rarewares.org/

LAME versions

Reply #48
Quote
3.97alpha5 is now available at Rarewares.
[a href="index.php?act=findpost&pid=263987"][{POST_SNAPBACK}][/a]


What's the difference between alpha 4 and alpha 5?

LAME versions

Reply #49
Quote
What's the difference between alpha 4 and alpha 5?
[a href="index.php?act=findpost&pid=264112"][{POST_SNAPBACK}][/a]

Gabriels post on the Lame 3.97.4 thread was:
Quote
3.97a5 is comitted in cvs.

I'd mainly like opinions about "-b 128 -X 10,10" against 3.90.3.

It also includes a few tuning changes to all presets.
John
----------------------------------------------------------------
My compiles and utilities are at http://www.rarewares.org/

 
SimplePortal 1.0.0 RC1 © 2008-2019