Lame 3.98b1
http://lame.bakerweb.biz/
...and now here
http://www.rarewares.org/mp3.html
Full Version: Lame LAME 3.98b1
I just tested it. Few samples only (herding_calls, lead_voice, trumpet, harp40_1, eig, trumpet: my prince) but spending more time than I did with my last very short a12 test.
I decided only to use -V1 cause from the recent alphas I know there was some improvement when going from -V2 to -V1, but next to none when going on to -V0.
trumpet: fine to me, I couldn't hear the very slight tremolo any more which I heard when testing a11.
harp40_1: fine to me as well.
herding_calls: still rather easily abxable but acceptable. (With this sample I did a comparison with a11 and couldn't abx the difference).
lead-voice: tremolo at the beginning still very easy to abx. Wouldn't call it annoying but I wouldn't call it easy to accept either.
trumpet my prince: Took me some time to get acquainted with the sample again, but after a while I could abx the slight tremolo again. However it's not serious so I would call it acceptable.
eig: as usual.
I decided only to use -V1 cause from the recent alphas I know there was some improvement when going from -V2 to -V1, but next to none when going on to -V0.
trumpet: fine to me, I couldn't hear the very slight tremolo any more which I heard when testing a11.
harp40_1: fine to me as well.
herding_calls: still rather easily abxable but acceptable. (With this sample I did a comparison with a11 and couldn't abx the difference).
lead-voice: tremolo at the beginning still very easy to abx. Wouldn't call it annoying but I wouldn't call it easy to accept either.
trumpet my prince: Took me some time to get acquainted with the sample again, but after a while I could abx the slight tremolo again. However it's not serious so I would call it acceptable.
eig: as usual.
I notice this was released just a few days after alpha12 so I'm guessing it's pretty much the same thing.
Finally a new lame beta. Best news in the changelog: Known problem samples for the new VBR code: many of them are at an acceptable quality level now; with a big 'Thank You' to Francis Niechcial.
Any chance we'll see an update in the List of recommended LAME settings?
Any chance we'll see an update in the List of recommended LAME settings?
QUOTE(twostar @ May 20 2007, 07:53) 
Any chance we'll see an update in the List of recommended LAME settings?
I guess they are the same. --vbr-new is now default, so you don't have to add it to command line.
QUOTE(Fishman0919 @ May 16 2007, 15:57)
I downloaded both the BakerWeb (847.872bytes 16/05/2007 15.01) and the Rarewares (577.536bytes 16/05/2007 22.54) builds as I intended to check whether there are differences in performances. Actually, after just testing CBR128 encoding/decoding cycle on two different systems I realized there is a lot of difference in decoding speed, so I decided to quit. I'm not much of an expert in LAME, there may even be a good explanation for that behaviour which I don't know about.
CODE
Prescott PIV 2.8ghz Orleans A64 3500+
Enc Dec Enc Dec
3.98 beta 1 - RareWares 12,6x 26,0x 15,2x 36,2x
3.98 beta 1 - BakerWeb 11,9x 66,5x 15,5x 115,0x
In this very short test session BakerWeb build decodes CBR128 2,5 times faster on the Prescott and over three times faster on the Orleans. I repeated the test also for 3.98alpha12 builds, with the same findings.
QUOTE(Josef Pohm @ May 21 2007, 06:50)
QUOTE(Fishman0919 @ May 16 2007, 15:57)
I downloaded both the BakerWeb (847.872bytes 16/05/2007 15.01) and the Rarewares (577.536bytes 16/05/2007 22.54) builds as I intended to check whether there are differences in performances. Actually, after just testing CBR128 encoding/decoding cycle on two different systems I realized there is a lot of difference in decoding speed, so I decided to quit. I'm not much of an expert in LAME, there may even be a good explanation for that behaviour which I don't know about.
CODE
Prescott PIV 2.8ghz Orleans A64 3500+
Enc Dec Enc Dec
3.98 beta 1 - RareWares 12,6x 26,0x 15,2x 36,2x
3.98 beta 1 - BakerWeb 11,9x 66,5x 15,5x 115,0x
In this very short test session BakerWeb build decodes CBR128 2,5 times faster on the Prescott and over three times faster on the Orleans. I repeated the test also for 3.98alpha12 builds, with the same findings.
Rarewares is compiled using ICL9.1
Bakerweb compile is MSVC8 i think...
Different compiles = different optimizations i think... plz correct me if i'm wrong!
btw: LAME 3.98b2 (325k) Monday, May 21, 2007 on Bakerweb
Very nice release, just --abr 288 makes a ~319kbps file??? i dunno why?
which quality is better 3.97 or 3.98b2 
(-V5 --vbr-new)
(-V5 --vbr-new)
Actually, I think the Bakerweb is an ICL4.5 compile, certainly the Intel compiler is specified and that particular Makefile is written for the 4.5 compiler.
Hello ! When will it be available for mac os X ???
QUOTE(Skylined ;)~ @ May 22 2007, 08:02)
btw: LAME 3.98b2 (325k) Monday, May 21, 2007 on Bakerweb
Very nice release, just --abr 288 makes a ~319kbps file??? i dunno why?
Well, that's not unusual, if your track demands many bits, it'll get them.
But I just found out that there is a bug in ABR's target bit allocation, which may result in corrupt mp3s. I think I fixed that, 3.98 beta 3 is in CVS. (CBR and VBR are not affected by this bug)
QUOTE(robert @ May 22 2007, 10:19)
....
But I just found out that there is a bug in ABR's target bit allocation, which may result in corrupt mp3s. I think I fixed that, 3.98 beta 3 is in CVS. (CBR and VBR are not affected by this bug)
Beta 3 now at Rarewares.
Has there been any fine tunings of the VBR presets since the last alpha?
I wonder what triggered the decision to label 3.98 as beta. This is now the third beta in a few days.
One would like to think that all usual settings and functions would be carefully tested during the alpha cycle. If I recall correctly, previously betas have been released when the encoder has been quite stable and suitable to be included in larger scale private and public listening tests. For example, each 3.97 beta was out for a few months.
Though, the name does not matter much. I am sure the LAME developers will let us know when 3.98 is ready for extensive testing and can be considered as a candidate for the "recommended version" approval.
Perhaps Sebastian Mares could reconsider starting the public 128 kbps MP3 test despite the previous poll results. A new LAME version changes the situation a bit.
One would like to think that all usual settings and functions would be carefully tested during the alpha cycle. If I recall correctly, previously betas have been released when the encoder has been quite stable and suitable to be included in larger scale private and public listening tests. For example, each 3.97 beta was out for a few months.
Though, the name does not matter much. I am sure the LAME developers will let us know when 3.98 is ready for extensive testing and can be considered as a candidate for the "recommended version" approval.
Perhaps Sebastian Mares could reconsider starting the public 128 kbps MP3 test despite the previous poll results. A new LAME version changes the situation a bit.
I second that
I labeled it beta, because the alpha version has so many bugs fixed, which are present in the 3.97 release version, that calling it still an 'alpha' version would be an understatement.
QUOTE(john33 @ May 22 2007, 08:57)
Actually, I think the Bakerweb is an ICL4.5 compile, certainly the Intel compiler is specified and that particular Makefile is written for the 4.5 compiler.
Bakerweb is ICL 6.0 compile.
Beta 3 is out, see rarewares 
What happened to beta 2 ?
What happened to beta 2 ?
QUOTE(le_canz @ May 22 2007, 12:06)
Beta 3 is out, see rarewares
What happened to beta 2 ?
JXL
I missed that....
QUOTE(robert @ May 22 2007, 13:19)
I labeled it beta, because the alpha version has so many bugs fixed, which are present in the 3.97 release version, that calling it still an 'alpha' version would be an understatement.
so do you think this release will become the HA recommended encoder before its stable then?
did the bitrate reservoir behavior was changed slighty from b1 to b3?
C:\Documents and Settings\All Users\Documents\Shared Programs\epsxe160\spuplayer
\wav>lameb -V5 "I Love TacticsOgre !!.wav"
LAME 3.98 (beta 1, May 16 2007) 32bits (http://www.mp3dev.org/)
CPU features: MMX (ASM used), 3DNow! (ASM used), SSE (ASM used)
Using polyphase lowpass filter, transition band: 15826 Hz - 16360 Hz
Encoding I Love TacticsOgre !!.wav to I Love TacticsOgre !!.wav.mp3
Encoding as 44.1 kHz VBR(q=5) j-stereo MPEG-1 Layer III (ca. 11.9x) qval=0
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
10376/10376 (100%)| 0:23/ 0:23| 0:23/ 0:23| 11.412x| 0:00
32 [ 203] ****
40 [ 0]
48 [ 0]
56 [ 0]
64 [ 16] *
80 [ 551] ***********
96 [ 789] %**************
112 [ 2305] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%********
128 [ 3552] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*********
160 [ 2959] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*
192 [ 0]
224 [ 1] %
256 [ 0]
320 [ 0]
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
126.6 75.3 24.7 99.2 0.5 0.4
Writing LAME Tag...done
ReplayGain: +9.1dB
CODE
C:\Documents and Settings\All Users\Documents\Shared Programs\epsxe160\spuplayer
\wav>lame -V5 "I Love TacticsOgre !!.wav"
LAME 3.98 (beta 3, May 22 2007) 32bits (http://www.mp3dev.org/)
CPU features: MMX (ASM used), 3DNow! (ASM used), SSE (ASM used)
Using polyphase lowpass filter, transition band: 15826 Hz - 16360 Hz
Encoding I Love TacticsOgre !!.wav to I Love TacticsOgre !!.wav.mp3
Encoding as 44.1 kHz VBR(q=5) j-stereo MPEG-1 Layer III (ca. 11.9x) qval=0
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
10376/10376 (100%)| 0:28/ 0:28| 0:28/ 0:28| 9.4586x| 0:00
32 [ 203] ****
40 [ 0]
48 [ 0]
56 [ 0]
64 [ 12] *
80 [ 497] **********
96 [ 833] %****************
112 [ 2287] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*********
128 [ 3477] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*********
160 [ 3066] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%**
192 [ 0]
224 [ 0]
256 [ 1] %
320 [ 0]
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
127.1 75.3 24.7 99.2 0.5 0.4
Writing LAME Tag...done
ReplayGain: +9.1dB
\wav>lame -V5 "I Love TacticsOgre !!.wav"
LAME 3.98 (beta 3, May 22 2007) 32bits (http://www.mp3dev.org/)
CPU features: MMX (ASM used), 3DNow! (ASM used), SSE (ASM used)
Using polyphase lowpass filter, transition band: 15826 Hz - 16360 Hz
Encoding I Love TacticsOgre !!.wav to I Love TacticsOgre !!.wav.mp3
Encoding as 44.1 kHz VBR(q=5) j-stereo MPEG-1 Layer III (ca. 11.9x) qval=0
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
10376/10376 (100%)| 0:28/ 0:28| 0:28/ 0:28| 9.4586x| 0:00
32 [ 203] ****
40 [ 0]
48 [ 0]
56 [ 0]
64 [ 12] *
80 [ 497] **********
96 [ 833] %****************
112 [ 2287] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*********
128 [ 3477] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*********
160 [ 3066] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%**
192 [ 0]
224 [ 0]
256 [ 1] %
320 [ 0]
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
127.1 75.3 24.7 99.2 0.5 0.4
Writing LAME Tag...done
ReplayGain: +9.1dB
CODE
C:\Documents and Settings\All Users\Documents\Shared Programs\epsxe160\spuplayer
\wav>lameb -V5 "I Love TacticsOgre !!.wav"
LAME 3.98 (beta 1, May 16 2007) 32bits (http://www.mp3dev.org/)
CPU features: MMX (ASM used), 3DNow! (ASM used), SSE (ASM used)
Using polyphase lowpass filter, transition band: 15826 Hz - 16360 Hz
Encoding I Love TacticsOgre !!.wav to I Love TacticsOgre !!.wav.mp3
Encoding as 44.1 kHz VBR(q=5) j-stereo MPEG-1 Layer III (ca. 11.9x) qval=0
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
10376/10376 (100%)| 0:23/ 0:23| 0:23/ 0:23| 11.412x| 0:00
32 [ 203] ****
40 [ 0]
48 [ 0]
56 [ 0]
64 [ 16] *
80 [ 551] ***********
96 [ 789] %**************
112 [ 2305] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%********
128 [ 3552] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*********
160 [ 2959] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*
192 [ 0]
224 [ 1] %
256 [ 0]
320 [ 0]
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
126.6 75.3 24.7 99.2 0.5 0.4
Writing LAME Tag...done
ReplayGain: +9.1dB
QUOTE(dannyb37 @ May 23 2007, 05:02)
QUOTE(robert @ May 22 2007, 13:19)
I labeled it beta, because the alpha version has so many bugs fixed, which are present in the 3.97 release version, that calling it still an 'alpha' version would be an understatement.
so do you think this release will become the HA recommended encoder before its stable then?I think Robert's explanation is reasonable. I just wondered if there are some other reasons that are not commonly known.
As I said, it is all about naming and what people expect from the beta label.
The discussion of 3.98 alphas has not been as lively as the discussion of 3.97 alphas was. Back then 3.90.3 was still HA's main recommendation and there was a bigger demand for a new version. LAME 3.97b2 was found to be good enough and it became the recommended version even though it was still beta. Most users have been satisfied with 3.97 so there is no similar demand now.
Looks like the development team has reached their main goals and now it is time to get feedback from a larger group of users. Naming it as beta is a way to increase popular interest. The best way we can help the developers is to try the LAME 3.98 betas and give that feedback so that they can do some final tweaking and the possibly remaining bugs can be ironed out.
Edit: grammar
QUOTE(Alex B @ May 23 2007, 08:45)
The discussion of 3.98 alphas has not been as lively as the discussion of 3.97 alphas was. Back then 3.90.3 was still HA's main recommendation and there was a bigger demand for a new version. LAME 3.97b2 was found to be good enough and it became the recommended version even it was not labelled as final yet. Most users have been satisfied with 3.97 so there is no similar demand now.
Looks like the development team has reached their main goals and now it is time to get feedback from a larger group of users. Naming it as beta is a way to increase popular interest. The best way we can help the developers is to try the LAME 3.98 betas and give that feedback so that they can do some final tweaking and the possibly remaining bugs can be ironed out.
Edit: grammar
When i remember right back in 3.97alpha days Guruboolez was the one that offered the most extensive listening and therefore often startet this "lively discusion"
Now with 3.98 builds he didnīt offer much findings in public anymore. Maybe cause of avoiding to waste to many time with debates on principles.
I bet non-public testing has already gone thru many many listening sessions
Regarding the quality improvements in 3.98b i am looking forward to the final release anyway. 3.97 had that many problems with added noise on many kinds of wind-instruments and voices i hardly had to listen much music to find some.
With 3.98b this is dramatically improved!
What's about MAC OS X version of LAME 3.98 ???? I cannot use it...
QUOTE(Wombat @ May 23 2007, 17:16)
When i remember right back in 3.97alpha days Guruboolez was the one that offered the most extensive listening and therefore often startet this "lively discusion"
Now with 3.98 builds he didnīt offer much findings in public anymore. Maybe cause of avoiding to waste to many time with debates on principles.
I bet non-public testing has already gone thru many many listening sessions
He did. 3.98a11 corresponds with 3.98a12 and 3.98b1 in terms of VBR quality.
IMO people on HA already satisfied with moden MP3 quality because 2006 multi-format listening test proved that MP3 can compete with other format. Since then, lower bitrate attracts more interest than 128kbps that is required to get well-encoded MP3.
QUOTE(dannyb37 @ May 23 2007, 04:02)
QUOTE(robert @ May 22 2007, 13:19)
I labeled it beta, because the alpha version has so many bugs fixed, which are present in the 3.97 release version, that calling it still an 'alpha' version would be an understatement.
so do you think this release will become the HA recommended encoder before its stable then?
Well, that's not my business. What version HA recommends is the decision of the community here. I hope the change invites people fearing 'alpha' versions to do some tests.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.