IPB

Welcome Guest ( Log In | Register )

2 Pages V   1 2 >  
Reply to this topicStart new topic
Lame 3.93.1, testing phase... (one week)
Gabriel
post Nov 23 2002, 19:51
Post #1


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



Lame 3.93.1 will be a corrected version of 3.93. (The new changes introduced by Takehiro will be in 3.94).
There is one week of testing period before releasing. You can download the candidate version here:
http://gabriel.mp3-tech.org/lame/lame_test.zip

I would like to hear both negative and positive reports.
If your results are positive, just drop a line, there is no need for a long post.
For negative reports, a more detailled post is needed, with at least a description an a pointer to the sample.

Reports about low bitrates abr presets and medium preset would be nice too.


Thank you for your particpation.

This post has been edited by Gabriel: Nov 24 2002, 13:23
Go to the top of the page
+Quote Post
JohnV
post Nov 24 2002, 21:18
Post #2





Group: Developer
Posts: 2797
Joined: 22-September 01
Member No.: 6



Ok, nice. smile.gif

I'll be doing some testing during next week. Hopefully all people who are capable will attend also.


--------------------
Juha Laaksonheimo
Go to the top of the page
+Quote Post
scott2002
post Nov 24 2002, 23:41
Post #3





Group: Members
Posts: 3
Joined: 24-November 02
Member No.: 3868



Howdy,
I am wondering why you don't release the corrected version of LAME as 3.94 and then have 3.95 with Takehiro's changes. I think this would be much better and help people forget 3.93 was released as a "new" version is released. It also helps stop the confusion for some newer people.
Cheers smile.gif
Scott
Go to the top of the page
+Quote Post
M
post Nov 24 2002, 23:53
Post #4





Group: Members
Posts: 964
Joined: 29-December 01
Member No.: 830



Well, that's how this thread was originally posted, if I remember correctly. My personal preference is for it to be done as 3.93.1 and 3.94, since many (I assume, but I'm one of 'em!) folks have already been experimenting with Takehiro's 3.94 alpha 2, and the introduction of a "new" 3.94 without the changes some of us have come to associate with that release could potentially cause as much confusion, if not more. No one knows how may copies of 3.94 alpha 2 may be floating around; therefore, ir might be best to keep all versions of 3.94 internally compatible.

- M.
Go to the top of the page
+Quote Post
scott2002
post Nov 25 2002, 00:06
Post #5





Group: Members
Posts: 3
Joined: 24-November 02
Member No.: 3868



Howdy,
I can see your point, but i'm sure you could get used to the idea that all of Takehiro changes could go into 3.95 and we have a "fixed 3.93" in 3.94 STABLE.

It would keep the versions smoother. Plus the "versions floating around" are only alphas.

Cheers,
Scott

This post has been edited by scott2002: Nov 25 2002, 00:08
Go to the top of the page
+Quote Post
frozenspeed
post Nov 25 2002, 01:12
Post #6





Group: Members
Posts: 207
Joined: 16-October 01
From: Seattle, WA
Member No.: 301



Has this fix been added to the cvs yet?

-Jeff
Go to the top of the page
+Quote Post
Gabriel
post Nov 25 2002, 08:43
Post #7


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



The fix is commited to cvs into the main branch.

For the version numbers, the decision is not stopped. The current opinion is to use 3.93.1 in order to avoid confision with the already floating-around 3.94 alphas.
Go to the top of the page
+Quote Post
scott2002
post Nov 25 2002, 11:52
Post #8





Group: Members
Posts: 3
Joined: 24-November 02
Member No.: 3868



Howdy,
Thats my point. They are only "alphas" not "releases" meaning they can change. I think we need lame 3.94 to contain the fix and correct the wrong....

In what way would it cause confusion? I could only see benifits. Lame 3.92 was released straight after 3.91 to correct the alt-fast standard bug if I remember correctly:)

What the point of having a 3.93.1 release?

Does anyone else think the same?

Scott smile.gif

This post has been edited by scott2002: Nov 25 2002, 11:53
Go to the top of the page
+Quote Post
Slo Mo Snail
post Nov 25 2002, 13:50
Post #9





Group: Members
Posts: 111
Joined: 2-July 02
From: Germany
Member No.: 2450



Where can one get the sources of lame 3.93.1? The zip file contains only one exe and in the lame cvs there's no 3.93.1 branch

EDIT: or is it the latest fix to lame/lame/libmp3lame/psymodel.c (1.114)?

This post has been edited by Slo Mo Snail: Nov 25 2002, 13:57
Go to the top of the page
+Quote Post
Gabriel
post Nov 25 2002, 14:08
Post #10


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



Yes, it is the latest commit to psymodel.c
Go to the top of the page
+Quote Post
ermannob
post Nov 25 2002, 15:04
Post #11





Group: Members
Posts: 14
Joined: 4-March 02
From: venice!
Member No.: 1441



Excuse me Gabriel,
what should i test? huh.gif
should i compare a file encoded with lame 3.93.1 and another compiled with lame 3.93?
always --alt-preset standard?

thanks
Go to the top of the page
+Quote Post
Gabriel
post Nov 25 2002, 16:55
Post #12


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



You should test the binary I provided against previous versions (3.92, 3.90.2,...). No need to test against 3.93, as this binary is made to correct 3.93.
Go to the top of the page
+Quote Post
Dibrom
post Nov 25 2002, 21:40
Post #13


Founder


Group: Admin
Posts: 2958
Joined: 26-August 02
From: Nottingham, UK
Member No.: 1



QUOTE (scott2002 @ Nov 25 2002 - 03:52 AM)
Howdy,
Thats my point. They are only "alphas" not "releases" meaning they can change.

There's already enough "room for change" in LAME releases. It's already confusing enough trying to estimate what will be in release to release. It has already been said by the LAME devs that Takehiro's code will be in 3.94. People are expecting this.

Are you going to explain to them that this is not the case each time someone asks what the deal with 3.94 is? I know that I don't want to, and trust me, there will be plenty of people asking these types of questions.

It makes more sense to release a bug fix as 3.93.1. Furthermore, I don't think a bug fix is so much indicative of an evolution or big progression in code or features so the revision number should be smaller accordingly.

Edit: I notice you are a new member on this board... so I don't know how long you've been reading, but if you go back and read some of the discussion about 3.94 and Takehiro's work, it might make more sense to you why 3.93.1 would be the better choice.

This post has been edited by Dibrom: Nov 25 2002, 21:41
Go to the top of the page
+Quote Post
HansHeijden
post Nov 25 2002, 23:44
Post #14





Group: Members
Posts: 159
Joined: 30-September 01
Member No.: 75



I just gave it another try, and still couldn't hear any difference between 3.92 and lame_test. Presets were --alt-preset cbr 128, standard, insane and also gpsycho at 128k.

I'll be away for work during a couple of days, hope to be back in the weekend.

Hans
Go to the top of the page
+Quote Post
R.A.F.
post Nov 26 2002, 11:44
Post #15





Group: Members
Posts: 281
Joined: 4-August 02
From: Nuremberg/Bavaria
Member No.: 2924



I fully agree with Scott2002. Why giving in 2 identical cases (bugfixes from 3.91 -> 3.92; now: 3.93 -> 3.93.1) two different version-syntaxes? It would be much easier to understand for people, who don't read HA each and every day, which version is the most actual. And over this an "outsider" for sure didn't hear anything about this 3.94a2-version, and also doesn't know where to download from, 'cause this version is still a little bit "hidden" on this HA-server. And for the others, who know about, it should be easily to explain, why 3.94a2 becomes in the near future a "3.95 stable".

- R.A.F. -


--------------------
My used codecs and settings:
FLAC V1.1.2 -4 / APE V3.99 Update 4 -high / MPC V1.15v --q 5 / LAME V3.97b2 -V2 --vbr-new / OGG aoTuV V4.51 Lancer -q5
Go to the top of the page
+Quote Post
SNYder
post Nov 26 2002, 20:53
Post #16





Group: Members
Posts: 317
Joined: 30-September 01
Member No.: 118



Does this release correct the problem with "--alt-preset fast" switches?
Go to the top of the page
+Quote Post
JohnV
post Nov 28 2002, 01:14
Post #17





Group: Developer
Posts: 2797
Joined: 22-September 01
Member No.: 6



-q0 is clearly worse in 3.93.1 cbr/abr than in 3.92 cbr/abr.
It's seriously worse especially with nspsytune.

Example (listen to the hihat hits):
Lame 3.92 --alt-preset cbr 128 -q0
t1_392_AltP-cbr128q0.mp3

Lame 3.93.1 --alt-preset cbr 128 -q0
t1_3931_AltP-cbr128q0.mp3

Gpsycho has less serious problem with this, but there's a quality difference also with gpsycho. vbr -q0 seems to be ok at least with this sample.

This post has been edited by JohnV: Nov 28 2002, 01:36


--------------------
Juha Laaksonheimo
Go to the top of the page
+Quote Post
Tex
post Nov 28 2002, 06:03
Post #18





Group: Members
Posts: 1
Joined: 28-November 02
Member No.: 3912



In listening to hihat.wav encoded with various releases of LAME including the 3.93.1 candidate, I've observed some differences that may be of some use. The differences have to do with extent to which the cow bell (I'm from Texas wink.gif ) sound at the end of hihat can be heard from the encoded file. The focus has been on --alt-preset 128 (--preset 128) and --alt-preset standard (--preset standard) for releases starting from 3.90.2.

The LAME releases are:
3.90.2 (dibrom)
3.92 (mitiok)
3.93 alphas posted at http://gabriel.mp3-tech.org/lame/
alpha2 dates identified as: 1May, 1Jul, 31Jul, 15Aug, 31Aug, 1Sep
alpha3 date identified as: 1Nov
3.93.1 the candidate posted as http://gabriel.mp3-tech.org/lame/lame_test.zip
and reported in the DOS window as 3.94 alpha1

--alt-preset 128 and --alt-preset standard are used for 3.90.2, 3.92 and the 3.93 alpha's through 15Aug.

--preset 128 and --preset standard are used for the 3.93 alphas after 15Aug and 3.93.1

Hihat.wav was obtained from http://lame.sourceforge.net/download/samples


Findings:
For --alt-preset 128 (--preset 128), only releases 3.90.2 and 3.93.1 reproduced the cow bell sound. All others truncated the sound partially or entirely.

For --alt-preset standard (--preset standard), the cow bell sound was truncated to the same extent for releases 3.92, 31Aug, and 3.93.1. All of the other releases seem to reproduce the cow bell sound.

The determinations were made by listening with modest equipment and comparing against the original .wav file. Nothing fancy.


Summary:
So far as the 3.93.1 candidate, truncation was observed for --preset standard. No truncations were observed with either setting for 3.90.2. Truncations were observed with both settings for 3.92.


Tex

P.S. - 28Nov - Further testing has revealed that the truncations coorelate with the player being used when the original testing was being performed, namely the Windows Media Player 7. With other players that don't have problems with ABR and VBR, no truncation occurs. Looks like the observations above are of no concern as regards the 3.93.1 candidate or the other releases.

This post has been edited by Tex: Nov 29 2002, 04:03
Go to the top of the page
+Quote Post
FinCoder
post Nov 28 2002, 07:51
Post #19





Group: Members
Posts: 31
Joined: 12-November 02
Member No.: 3759



Took some samples from:

http://lame.sourceforge.net/download/samples

I am not done yet, but I noticed something...

I used command line like this:

lame ftb_samp.wav --alt-preset cbr 128 -q0

lame-394-alpha2: 14 seconds
lame3.90.2-ICL: 1 minute and 37 seconds to finnish!!!
lame3.90.2-MSVC: 1 minute 38 seconds!!
lame test :12 seconds and same result with binary (3.93) from this page

http://irgendwas.mybinaryblocks.com/~mitiok/

btw. all other compiles allows you to use lower bitrates than other 2 compiles...if you try to run ICL or MSVC compile like this:lame name-of-the-file --alt-preset cbr 64 -q0 no way, the lowest bitrate is 80, when it works.

I quess this does not help a bit...but i will test a litle bit more

This post has been edited by FinCoder: Nov 28 2002, 08:22
Go to the top of the page
+Quote Post
Gabriel
post Nov 28 2002, 11:18
Post #20


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



@JohnV: do you think we should put back -q0 to be identical as -q1?
Go to the top of the page
+Quote Post
Gabriel
post Nov 28 2002, 11:20
Post #21


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



No, the bitrate problem of --preset fast standard is not corrected in this release.
Go to the top of the page
+Quote Post
JohnV
post Nov 28 2002, 11:35
Post #22





Group: Developer
Posts: 2797
Joined: 22-September 01
Member No.: 6



QUOTE (Gabriel @ Nov 28 2002 - 12:18 PM)
@JohnV: do you think we should put back -q0 to be identical as -q1?

Well, if you listened those samples, you probably agree that you should do something..
If fixing -q0 with cbr/abr is not an option/requires too much work, then you should do something else..


--------------------
Juha Laaksonheimo
Go to the top of the page
+Quote Post
Gabriel
post Nov 28 2002, 11:44
Post #23


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



I think that fixing -q0 is not feasible in a short time. Previously it was the same as -q1, only in november it was changed.
But if I put it back to the q1 state, it will also be for vbr.
Go to the top of the page
+Quote Post
JohnV
post Nov 28 2002, 12:08
Post #24





Group: Developer
Posts: 2797
Joined: 22-September 01
Member No.: 6



QUOTE (Gabriel @ Nov 28 2002 - 12:44 PM)
I think that fixing -q0 is not feasible in a short time. Previously it was the same as -q1, only in november it was changed.
But if I put it back to the q1 state, it will also be for vbr.

Imo it would be wise to go back then.. Some people religiously think that -q0 gives always the best results..


--------------------
Juha Laaksonheimo
Go to the top of the page
+Quote Post
Gabriel
post Nov 28 2002, 12:14
Post #25


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



Agree...
Go to the top of the page
+Quote Post

2 Pages V   1 2 >
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 22nd December 2014 - 20:33