Skip to main content

Topic: LAME 3.97 Stable (Read 21434 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • ChiGung
  • [*][*][*][*]
LAME 3.97 Stable
Reply #25
fwiw, this IChing looks auspicious...
Quote
The answer to the question, "should i expect lame 397 before august" is:

May 7, 105
19. Lin - Approach
          -- --
          -- --     above     K'un   The Receptive, Earth
          -- --
          -- --
          -----     below     Tui    The Joyous, Lake
          -----

     The Judgement
          Approach has supreme success.
          Perseverance furthers.
          When the eighth month comes,
          There will be misfortune.

     The Image
          The earth above the lake:
          The image of Approach.
          Thus the superior man is inexhaustible
          In his will to teach,
          And without limits
          In his tolerance and protection of the people.

     Changing Lines
          Changing yang at the bottom means:
          Joint approach. Perseverance brings good fortune.

          Changing yang in the second place means:
          Joint approach. Good fortune. Everything furthers.

          Changing yin in the third place means:
          Comfortable approach. Nothing that would further.
          If one is induced to grieve over it, One becomes free of blame.

          Changing yin in the fourth place means:
          Complete approach. No blame.

          Changing yin in the fifth place means:
          Wise approach. This is right for a great prince.Good fortune.

          Changing yin at the top means:
          Greathearted approach.
          Good fortune. No blame.
no conscience > no custom

  • music_man_mpc
  • [*][*][*][*][*]
  • Members (Donating)
LAME 3.97 Stable
Reply #26
Quote
when should we expect LAME 4.0?
[a href="index.php?act=findpost&pid=295905"][{POST_SNAPBACK}][/a]

Probably anytime between a month or two from now and next year.  Beta testing should be coming along *fairly soon*, but I'm not sure exactly how long *fairly soon* is and I have absolutely no idea how long it will be in beta stage before we see a stable release.
Quote
Any quality differences besides the speed?
[a href="index.php?act=findpost&pid=295905"][{POST_SNAPBACK}][/a]

Yes, the quality will be initially worse than recent 3.9X releases but LAME 4.0 should be capable of better quality than 3.9X in the long run.  However it will probably need to be tuned considerably first I don't think anyone could reasonably estimate how long said tuning will take.
Quote
When should we expect the LAME 3.97?
[a href="index.php?act=findpost&pid=295905"][{POST_SNAPBACK}][/a]

Also *fairly soon.*  It could be released tomorrow, it could take another month or two.  Your guess is as good as anyone's at this point.
Quote
And when should we expect HO.org to update their recommended encoder?
[a href="index.php?act=findpost&pid=295905"][{POST_SNAPBACK}][/a]

As soon as 3.97 stable is released, this has been stated a number of times.
gentoo ~amd64 + layman | ncmpcpp/mpd | wavpack + vorbis + lame

  • Busemann
  • [*][*][*][*][*]
LAME 3.97 Stable
Reply #27
Quote
As soon as 3.97 stable is released, this has been stated a number of times.
[a href="index.php?act=findpost&pid=295914"][{POST_SNAPBACK}][/a]


I might have missed the obvious, but why will 3.97 be recommended without any testing, while 3.96.1 which went through lots of thorough testing still isn't?

  • guruboolez
  • [*][*][*][*][*]
  • Members (Donating)
LAME 3.97 Stable
Reply #28
I've counted the number of ABX tests I performed during january-march for 3.97 alphas : ~800. It corresponds to more than 10.000 ABX trials. Just look on the MP3-TECH forum.
Other people also posted positive tests for 3.97 alphas.

  • Gen912
  • [*]
LAME 3.97 Stable
Reply #29
Quote
...especially when the question you are asking has already been answered in the thread that is linked to by an earlier post in this one.
Quote

Gabriel's post confirms that.
[a href="index.php?act=findpost&pid=295679"][{POST_SNAPBACK}][/a]

[a href="index.php?act=findpost&pid=295849"][{POST_SNAPBACK}][/a]


Ok, sorry. I must admit that I didn't read the post Jaybeee linked to before answering rjamorim.

Gabriels post he linked to, plus post #19 by Jean Luc and your post #20 answer my question below.


Quote
Quote

Quote

Ok, let me rephrase my question:
Why is there work being done on  a 4.x version, when there isn't a stable/final release of 3.97 yet?
Does this have to do with different teams of developers?
[a href="index.php?act=findpost&pid=295789"][{POST_SNAPBACK}][/a]

To make things simple: Takehiro works on 4.0, Gabriel and Robert work on 3.x
So, yeah, you could consider them "different teams"
[a href="index.php?act=findpost&pid=295790"][{POST_SNAPBACK}][/a]

...why isn't Takehiro working on 3.97 together with Gabriel and Robert?
Or vice versa...., why not drop further developement of 3.97 and concentrate all efforts on 4.x?
[a href="index.php?act=findpost&pid=295798"][{POST_SNAPBACK}][/a]


In conclusion; I'm not in a hurry for a new version. I prefer developers taking their time to deliver quality than them getting rushed into delivering something nobody's really happy with.

Succes wished to the developers.
I have no particular talent, I am merely inquisitive.  Albert Einstein

  • schonenberg
  • [*][*][*]
  • Banned
LAME 3.97 Stable
Reply #30
Quote
Maybe I'm getting on your nerves now (don't mean to) but, why isn't Takehiro working on 3.97 together with Gabriel and Robert?
Or vice versa...., why not drop further developement of 3.97 and concentrate all efforts on 4.x?
[{POST_SNAPBACK}][/a]


Why not read takehiro's posts:
[a href="http://www.hydrogenaudio.org/forums/index.php?act=Search&CODE=getalluser&mid=2067]http://www.hydrogenaudio.org/forums/index....lluser&mid=2067[/url]

read the LAME4 TODO:
http://cvs.sourceforge.net/viewcvs.py/lame...07-experimental
  • Last Edit: 08 May, 2005, 12:27:41 AM by schonenberg

  • ShowsOn
  • [*][*][*][*]
  • Members (Donating)
LAME 3.97 Stable
Reply #31
I read part of the LAME4 Todo list, one of the points states this:

"90.
Use intensity stereo. This is a must-have for low bitrates, but if the
algorythm is very good it could also be used in every case."

Does that mean that a high quality VBR file (say -V 2) could use intensity stereo on low bitrate frames, in order improve the quality of those frames to keep the bit rate even lower?

The last to-do list item is to make LAME as fast as fast as FASTENC, exactly how fast is FASTENC compared with the LAME4 alphas?

How does the quality of FASTENC compare with 3.9x and 4 alpha?

  • alfa156
  • [*][*][*]
LAME 3.97 Stable
Reply #32
Quote
Quote
As soon as 3.97 stable is released, this has been stated a number of times.
[a href="index.php?act=findpost&pid=295914"][{POST_SNAPBACK}][/a]


I might have missed the obvious, but why will 3.97 be recommended without any testing, while 3.96.1 which went through lots of thorough testing still isn't?
[a href="index.php?act=findpost&pid=295922"][{POST_SNAPBACK}][/a]


quality tests included of course...

  • ChiGung
  • [*][*][*][*]
LAME 3.97 Stable
Reply #33
Quote
I've counted the number of ABX tests I performed during january-march for 3.97 alphas : ~800. It corresponds to more than 10.000 ABX trials. Just look on the MP3-TECH forum.
Other people also posted positive tests for 3.97 alphas.

That is quite amazing - you deserve honours for that.
I was thinking the only way to get that volume of testing done, would
be to make an applet of the java abcx'er and collect surf-by responses.
To think that much testing has been done, just recently, by one person..

'more power to your elbow' guruboolez
no conscience > no custom

  • Gabriel
  • [*][*][*][*][*]
  • Developer
LAME 3.97 Stable
Reply #34
Some things can grab a lot of time, like job or moving to a new apartment. Some things can be problematic for developement, like DSL connection not yet restored due to recent moving...

  • rjamorim
  • [*][*][*][*][*]
LAME 3.97 Stable
Reply #35
Quote
"90.
Use intensity stereo. This is a must-have for low bitrates, but if the
algorythm is very good it could also be used in every case."

Does that mean that a high quality VBR file (say -V 2) could use intensity stereo on low bitrate frames, in order improve the quality of those frames to keep the bit rate even lower?


It means a high quality VBR file could use intensity stereo in ALL frames.

Quote
The last to-do list item is to make LAME as fast as fast as FASTENC, exactly how fast is FASTENC compared with the LAME4 alphas?

How does the quality of FASTENC compare with 3.9x and 4 alpha?[a href="index.php?act=findpost&pid=296002"][{POST_SNAPBACK}][/a]


It's nothing but a disservice to start doing these comparisions now, while 4 is barely past alpha stage.
Get up-to-date binaries of Lame, AAC, Vorbis and much more at RareWares:
http://www.rarewares.org

  • Madrigal
  • [*][*][*][*]
LAME 3.97 Stable
Reply #36
Quote
Some things can grab a lot of time, like job or moving to a new apartment. Some things can be problematic for developement, like DSL connection not yet restored due to recent moving...
[a href="index.php?act=findpost&pid=296077"][{POST_SNAPBACK}][/a]
For the record, Gabriel, by starting this thread I certainly didn't mean to criticise or complain about your efforts. I realize you have a life away from LAME, and that development is a big, time-consuming, and sometimes apparently thankless job. I had no idea you were in the middle of a move -- had I known, I would probably not have started this thread.

Thank you from the bottom of my heart for your labor of love, and all the hours of audio pleasure it has given me. I look forward to the day when we can all celebrate the release of a stable 3.97.

Once again, many thanks.

Regards,
Madrigal

  • callmeace
  • [*][*]
LAME 3.97 Stable
Reply #37
Quote
........Thank you from the bottom of my heart for your labor of love, and all the hours of audio pleasure it has given me......


That goes for me too Gabriel and all others that have put a lot of time and effort into the development of LAME. I've used it quite a bit directly - and I know LAME has been implememnted in other software which I have used also. I certainly appreciate it 

  • IgorC
  • [*][*][*][*][*]
LAME 3.97 Stable
Reply #38
3.90 3.96 3.97 4.xx 
I´m not trying to say that LAME is dead. However is there really big difference between 3.90 and 3.97 or 4.00? May be here some optimization, tuning , etc. , rebuliding algoritms. But when I try to go a bit less bitrate  VBR 121 kbit/s of 3.96 or
3.97 is getting worse(or in the best case the quality is igual) than 128 kbits 3.90. 
So gain is small 128/121 = 1,057..... (5-6% for 4-5 years of LAME´s develompent) (for me 3.90.3 is still best)
As I understand future new versions of LAME will provide more stability, optimization,speed, tuning etc, but quality gain will be smallest. However there is cense to keep development of LAME , since MP3 is very popular audiocodec and wisely supported
  • Last Edit: 09 May, 2005, 09:18:19 PM by IgorC

  • beto
  • [*][*][*][*][*]
  • Members (Donating)
LAME 3.97 Stable
Reply #39
I really think you are trolling, but I'll give you another chance.

LAME 3.97 (3.96.1) is faster than the 3.90 branch. Recently guruboolez made several ABX tests and it was decided that 3.97, once stable, will be the new recommended version. There were improvements qualitywise (or, better saying there were no significant regressions).

As for LAME4 it is expected to be faster than the 3.9X series, but with lower quality in a first moment. This will be reverted when proper tuning is done.

Next time do some ABX testing before posting claims about quality like you did.
  • Last Edit: 09 May, 2005, 09:41:28 PM by beto

  • ShowsOn
  • [*][*][*][*]
  • Members (Donating)
LAME 3.97 Stable
Reply #40
Quote
It's nothing but a disservice to start doing these comparisions now, while 4 is barely past alpha stage

I wasn't expecting a detailed comparison, just a rough idea if that last goal on the todo list is close, or still a long way off.

Each stable version of LAME seems to be getting faster, where as FASTENC is a stationary a  target. I've never used FASTENC so I have no idea exactly how fast it is.


What is the benefit of using intensity stereo on all frames? Is there somewhere that I can read more about the advantages of intensity stereo?
  • Last Edit: 09 May, 2005, 09:58:23 PM by ShowsOn

  • westgroveg
  • [*][*][*][*][*]
LAME 3.97 Stable
Reply #41
4.0 is a complete re-write of the LAME code & because they have the 3.x code to look at for mistakes & possible improvements I don't see why anyone would think 4.0 will produce lower quality audio than 3.x. I think 4.0 will be a HUGE improvement over yearly versions of LAME & make development & tweaks much easier.

  • beto
  • [*][*][*][*][*]
  • Members (Donating)
LAME 3.97 Stable
Reply #42
Quote
4.0 is a complete re-write of the LAME code & because they have the 3.x code to look at for mistakes & possible improvements I don't see why anyone would think 4.0 will produce lower quality audio than 3.x. I think 4.0 will be a HUGE improvement over yearly versions of LAME & make development & tweaks much easier.
[{POST_SNAPBACK}][/a]


I guess you misread my post. Of course it will be an improvement. But the fact of it being a complete rewrite of the code is likely to introduce new bugs IN A FIRST MOMENT.
You can read [a href="http://www.hydrogenaudio.org/forums/index.php?showtopic=29575&view=findpost&p=257995]here[/url] to see that the main goal at this moment is not quality but remove obstacles on tweaking.

  • IgorC
  • [*][*][*][*][*]
LAME 3.97 Stable
Reply #43
Quote
I really think you are trolling, but I'll give you another chance.

LAME 3.97 (3.96.1) is faster than the 3.90 branch. Recently guruboolez made several ABX tests and it was decided that 3.97, once stable, will be the new recommended version. There were improvements qualitywise (or, better saying there were no significant regressions).

As for LAME4 it is expected to be faster than the 3.9X series, but with lower quality in a first moment. This will be reverted when proper tuning is done.

Next time do some ABX testing before posting claims about quality like you did.
[a href="index.php?act=findpost&pid=296450"][{POST_SNAPBACK}][/a]


I don´t care about speed. 3.90.3 is still best for my ears. 3.96 seems to cut a bit much high freq. Sometimes I´ve tested 3.97  alpha 5 .... 10 , I liked alpha 5 and 6.
It´s not about professional comparison or anything like that (not a numeric test).
Just IMHO, not your  . Here you can´t say me how to listen.

  • Danimal
  • [*][*][*]
LAME 3.97 Stable
Reply #44
Quote
Quote
I really think you are trolling, but I'll give you another chance.

LAME 3.97 (3.96.1) is faster than the 3.90 branch. Recently guruboolez made several ABX tests and it was decided that 3.97, once stable, will be the new recommended version. There were improvements qualitywise (or, better saying there were no significant regressions).

As for LAME4 it is expected to be faster than the 3.9X series, but with lower quality in a first moment. This will be reverted when proper tuning is done.

Next time do some ABX testing before posting claims about quality like you did.
[a href="index.php?act=findpost&pid=296450"][{POST_SNAPBACK}][/a]


I don´t care about speed. 3.90.3 is still best for my ears. 3.96 seems to cut a bit much high freq. Sometimes I´ve tested 3.97  alpha 5 .... 10 , I liked alpha 5 and 6.
It´s not about professional comparison or anything like that (not a numeric test).
Just IMHO, not your  . Here you can´t say me how to listen.
[a href="index.php?act=findpost&pid=296456"][{POST_SNAPBACK}][/a]


He may not be able to tell you how to listen but the terms of service are quite specific as to what is required when you are talking here about what you claim you're hearing.  Click on terms of service at the top and look at no. 8.

If 3.96 "Seems to cut a bit much high freq." as you claim, then you ought to be able to pick that out when doing a double blind test and post your abx results here.  If you can't, then you are talking about placebo effect (you think that 3.96 cuts a bit too much high freq. because you either expect it to, or for any reason want it to).  If you can identify this problem in a double blind test then you have found something that would be of value for the LAME devs to hear more about.  If not, well the internet is full of places where uninformed speculation and opinions are the norm and welcomed with open arms.  This particular forum isn't one of them.

The entire reason HA was started was that the leader of another forum had developed his own set of mp3 tunings and refused to stop recommending them even after the alt presets were shown to be better.

  • ChiGung
  • [*][*][*][*]
LAME 3.97 Stable
Reply #45
Quote
If 3.96 "Seems to cut a bit much high freq." as you claim, then you ought to be able to pick that out when doing a double blind test and post your abx results here.

Agreed hes talking nonsense, just doesnt realise yet, that doesnt go down well here

About 397s release - I think it should be held back until 4.xx release. That way we can be sure to get the most tweaks Gabriel has left in him for it, and there will be a clear recommended version for 4xx to work towards.
Release 4xx after and lots of mistaken people will be 'upgrading' to it. -embrace their confusion release 4xx first 
  • Last Edit: 10 May, 2005, 12:23:06 AM by ChiGung
no conscience > no custom

  • music_man_mpc
  • [*][*][*][*][*]
  • Members (Donating)
LAME 3.97 Stable
Reply #46
Quote
Release 4xx after and lots of mistaken people will be 'upgrading' to it. -embrace their confusion release 4xx first 
[a href="index.php?act=findpost&pid=296470"][{POST_SNAPBACK}][/a]

This sounds like a good point to me, I wonder how much longer LAME4 stable will take then LAME 3.97?  Certainly if they're releases are less then a month apart this may well be worth looking into for the developers but I don't think 3.97 should be held back for any undue amount of time just for the sake of n00b confusion.
gentoo ~amd64 + layman | ncmpcpp/mpd | wavpack + vorbis + lame

  • Madrigal
  • [*][*][*][*]
LAME 3.97 Stable
Reply #47
Quote
About 397s release - I think it should be held back until 4.xx release. That way we can be sure to get the most tweaks Gabriel has left in him for it, and there will be a clear recommended version for 4xx to work towards.
If I understand it correctly, the release of 3.97 has already been held back significantly, in order to yield a "clear recommended version" -- to that end, all needed tweaks will certainly be applied beforehand.

But to hold back 3.97 until 4.xx is ready for release seems just plain silly to me. 3.97 is slated to have a significant role in the ongoing saga of LAME, before the release of 4.xx, and the former should be released as soon as humanly possible so that we can all enjoy the benefits of it.

EDIT: @music_man_mpc -- Amen. 3.97 should NOT be held back for any undue amount of time, just for the sake of n00b confusion.

Regards,
Madrigal
  • Last Edit: 10 May, 2005, 11:02:19 AM by Madrigal

  • IgorC
  • [*][*][*][*][*]
LAME 3.97 Stable
Reply #48
I´m not tending to troll. I read here  http://www.hydrogenaudio.org/forums/index....topic=30547&hl=. I  analyzed the results of 3.90 , 3.96, 3.97alpha 5
There wasn´t  single result  which has shown  wich is the best version. Also there are differnet settings play with it(q0, ABR,VBR etc even MP3gain) and different points of view how to test.
I´ll try to do some ABX test this week with a lot of samples not for  change somebody ´s opinion , but maybe mine own.

Where is 397alpha5? On rarewares there´s only last alpha10.
  • Last Edit: 10 May, 2005, 12:58:39 PM by IgorC

  • Madrigal
  • [*][*][*][*]
LAME 3.97 Stable
Reply #49
If you're not intending to troll, then you seem to be doing a pretty good job of it without intending to.

Have a nice day.

Regards,
Madrigal