Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Lame's future after --alt-presets (Read 25979 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Lame's future after --alt-presets

Reply #25
Yeah, this really is bad news...
I must say that according to all i've read in this forum about how LAME development is handled, and how quality doesn't matter to those who make calls, is really frustrating, and can certainly understand why Dibrom's fed up with all that... It's just too plain frustrating.
Well, what i can offer is my listening, since i don't have coding talents. Whenever "important scale" testing will be needed i'll be glad to help.

Lame's future after --alt-presets

Reply #26
Quote
I tested how close Lame 3.90.2 --alt-preset standard I can get with quick external switch tweaking using Lame 3.94a2. (Mitiok does not host alpha2)

Pls use 3.94alpha4.

alpha2/3 contains fatal bug which may make "invalid" mp3 file.
alpha2 contains stupid bug in block type switching (I made a typo 1.0 from 0.1).

So I think you should test with at least alpha3, especially if you want to check the pre-echo related problem.

Quote
-X0: noise measurement type selected based on quick testing over -X1

I have not tested much, but it seems -X0 is needed for short blocks and -X1 is needed for long blocks (in ABR/CBR mode). I wonder why this is happen.... Maybe because of short block psymodel mess.

In the old-vbr mode, -X0 and -X1 makes difference only when the 320kbps is not enough. In the such case, the block type is a short type almost always.

So the -X0 for vbr preset will be Ok I think.
May the source be with you! // Takehiro TOMINAGA

Lame's future after --alt-presets

Reply #27
Quote
alpha2 contains stupid bug in block type switching (I made a typo 1.0 from 0.1).

So I think you should test with at least alpha3, especially if you want to check the pre-echo related problem.

I'd hope mitiok would start providing latest alphas instead of giving the alpha1 as "dailylame".
I'd hope somebody would clear the situation what's going on in Lame CVS. Is your alpha4 the one which is gonna become 3.94. Hopefully, but I'm just wondering because Mitiok provides dailylame 3.94 alpha1..
Juha Laaksonheimo

Lame's future after --alt-presets

Reply #28
I can explain the situation:
The main branch, after 3.93 as been flagged as 3.94. So if you retrieve main branch, you'll have something that is said to be 3.94 alpha1. But as you know about the 3.93 problem, we will have 3.93.1. That is what is in the current main branch but still flagged as 3.94alpha1.

The "real" 3.94 is in Takehiro's branch. I'll try to post a compile of this thing this evening.

(note: perhaps this is an indication that we should have a branch policy)

Lame's future after --alt-presets

Reply #29
Which branch of Takehiro is the real 3.94 and does this branch include to bugfix from 3.93.1?

-Jeff

Lame's future after --alt-presets

Reply #30
I have encoded some samples with 394a2 and have some "unqualified"
meaning so far. What ever they did, lame has become more punchy.
I mean where some might say this is softened and an abx war starts.
Seems to be the new block switching.
I used JohnV´s command line that is the consequent way to go with only
switches. Nothing to add. Now lets find samples that don´t work with a4

Wombat
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

Lame's future after --alt-presets

Reply #31
So I gather that the forthcoming 3.94alpha4 compile will represent the most advanced/tweaked LAME code to date?

Lame's future after --alt-presets

Reply #32
Where could I get 3.94alpha4 ?

Lame's future after --alt-presets

Reply #33
It is very sad news that Dibrom has left LAME and he deserves all of our appreciation.

He seemed to be the only one who was not too 'close' to the technical details to remember the real value of MP3 - outstanding, easily transferrable audio quality in a SMALL sized file. Since the majority of people who use MP3s use them on portable devices (if you are at home, just listen to a CD or WAVE file, why mess with MP3???? What value doe sit provide when the file size is not a constraint), a small sized file that provide soutstanding quality is most desired. it seems the other programmers place no emphasis on the file size part of the equation, only on the quality size.

Again, last time I checked WAV and APE files provided PERFECT quality. Why not just use those???

Maybe some day someone will develop a 'portable' setting. until then, we have APS -Y if you have a digital jukebox and aps CBR 128 if you have a flash player.

Dibrom, thanks. we will miss your work.

Lame's future after --alt-presets

Reply #34
Wav and ape eats too much space.

Lame's future after --alt-presets

Reply #35
Quote
3.94a2 APS seems to have somewhat higher bitrates than 3.90.2 or 3.92 but it isn't enough to cause alarm. EncSpot reveals that 3.94a2 employs more short blocks than the previous versions, in some cases much more. I have several files where short blocks make up 15-20% of all blocks.

I really noticed the 3.94a2's short block handling on Paul Simon's "Me And Julio Down By The School Yard" with nearly 32% shortblocks, compared to 3.92's 16%. I guess that's from:

Quote
takehiro: alpha2 contains stupid bug in block type switching (I made a typo 1.0 from 0.1).


I was wondering if there was a LAME switch that can be used to adjust the robustness/sensitivity of the blockswitching or if that's something that might be useful later on to use with some highly tuned VBR command line such as what JohnV is going for?

Lame's future after --alt-presets

Reply #36
Could someone please compile a binary of LAME 3.94 alpha4? I tried looking through http://static.hydrogenaudio.org/extra/LAME/ but I couldn't find it there. I'm guessing this is the best version to use if I want to stick with bitrates in the range of 112-160 kbps (and MP3 is a must)?

Lame's future after --alt-presets

Reply #37
Quote
Could someone please compile a binary of LAME 3.94 alpha4? I tried looking through http://static.hydrogenaudio.org/extra/LAME/ but I couldn't find it there. I'm guessing this is the best version to use if I want to stick with bitrates in the range of 112-160 kbps (and MP3 is a must)?

I would happily do so, but I have no idea where to obtain the source from!! I presume it's not in the standard CVS. Is it in Takehiro's CVS branch?

Lame's future after --alt-presets

Reply #38
Quote
I can explain the situation:
The main branch, after 3.93 as been flagged as 3.94. So if you retrieve main branch, you'll have something that is said to be 3.94 alpha1. But as you know about the 3.93 problem, we will have 3.93.1. That is what is in the current main branch but still flagged as 3.94alpha1.

The "real" 3.94 is in Takehiro's branch. I'll try to post a compile of this thing this evening.

...

I guess you're right John33 (according to what Gabriel wrote).

Lame's future after --alt-presets

Reply #39
It seems to me it's still more than just a "little" early to start using 3.94alpha4 for anything other than testing...

Some of the comments here seem to indicate other purposes.

And also:
@ChS: if you encoded your file with alpha2, it might be useful to try to repeat the encode with alpha4 and see the % of short-blocks.

Unfortunately, getting an alpha4 binary doesn't seem to be easy...

Lame's future after --alt-presets

Reply #40
Quote
I think this is partially why my own approach didn't work out in the end.

I think this is true. Politics is neccesary in this endeavor.

And the sad fact is this:
Sound quality isn't really the priority with the LAME developers or they would have embraced Dib's efforts.

Lame's future after --alt-presets

Reply #41
I am wondering how are some of you able to know what are priorities of the few Lame developpers....

Some of you are assuming that is is not sound quality. Well, so tell me how Lame managed to produced not-so-bad mp3 files. It should be worst than Xing or Blade if quality is not a priority, shouldn't it?

You just have to realize that what most of you want is just a new front-end, nothing else. That is not a big deal, you can easily make it and link with libmp3lame.
Until the end of Lame 3.x, the default front-end will keep compatibility with older versions. Only with V4 it could change. In the meantime, it seems reasonable to do a new front-end if you need it.

Lame's future after --alt-presets

Reply #42
You can download a copy of Takehiro's 3.94 alpha 4 here. 

Lame's future after --alt-presets

Reply #43
Quote
I am wondering how are some of you able to know what are priorities of the few Lame developpers....

Some of you are assuming that is is not sound quality. Well, so tell me how Lame managed to produced not-so-bad mp3 files. It should be worst than Xing or Blade if quality is not a priority, shouldn't it?

You just have to realize that what most of you want is just a new front-end, nothing else. That is not a big deal, you can easily make it and link with libmp3lame.
Until the end of Lame 3.x, the default front-end will keep compatibility with older versions. Only with V4 it could change. In the meantime, it seems reasonable to do a new front-end if you need it.

I'll tell you how. The BEST thing that EVER happened to Lame sound quality was the alt presets. How was Dibrom treated by you guys? Were his efforts applauded or even appreciated? Did you guys invite him to join the developement team?

Who wants "not-so-bad mp3 files" anyway? NOT ME. I want the BEST. If you guys really want the best TOO, if sound quality is INDEED the priority, Dibrom would BE a Lame developer instead of just an outcast. An encoder which isn't fine tuned is WORTHLESS as far as I'm concerned.

This has NOTHING to do with frontends. I can write whatever frontend people want. John Edwards can write an even better one. It's not about frontends, it's about sound quality PERIOD.

Lame's future after --alt-presets

Reply #44
Quote
And also:
@ChS: if you encoded your file with alpha2, it might be useful to try to repeat the encode with alpha4 and see the % of short-blocks.

Checked it out with john33's alpha4 compile and the short block % was exactly the same as alpha2 at almost 32% (31.5% to be exact, which alpha2 was). To make it clear, that's with --preset standard.

Lame's future after --alt-presets

Reply #45
Quote
Checked it out with john33's alpha4 compile and the short block % was exactly the same as alpha2 at almost 32% (31.5% to be exact, which alpha2 was). To make it clear, that's with --preset standard.

There's lots of short block tunings and possibility in general to much better transient handling with 3.94. So obviously Dibrom should compensate this in --preset standard..
Juha Laaksonheimo

Lame's future after --alt-presets

Reply #46
Quote
Did you guys invite him (Dibrom) to join the developement team?

yes, Dibrom has every possibility as a LAME developer, since long ago. He just doesn't make use of it.

Lame's future after --alt-presets

Reply #47
Quote
You just have to realize that what most of you want is just a new front-end, nothing else. That is not a big deal, you can easily make it and link with libmp3lame.

No.

What I want is fundamental improvements to the psymodel.  I want consolidation and clarity in the code.  I want stability, and I want quality assurance.

Most of all, I want a fundamentally different approach towards project development and management.

None of this is related to a frontend.

Lame's future after --alt-presets

Reply #48
Quote
Quote
Did you guys invite him (Dibrom) to join the developement team?

yes, Dibrom has every possibility as a LAME developer, since long ago. He just doesn't make use of it.

Yes, I have the status of a LAME developer, but only as a singular entity.  What I don't have is sway with the rest of the developers.  I've made no secret of the fact that the alt-presets have hit a ceiling and cannot get better without help from one of the other developers -- one of the developers who knows the psymodel well.  Recently, this has become possible since Takehiro has started to show some interest.  However, the problem now is that I can't convince the rest of the developers to adopt a new approach or to change many of the things in LAME that I think need to be changed.  So, yes, I have the "title" of a LAME developer, but for practical purposes (at least for me) it is meaningless.

Lame's future after --alt-presets

Reply #49
If you needed help, why didn't you ask for help? You know e-mail adresses of LAME devs, there are mailing lists...