http://gabriel.mp3-tech.org/lame/lame_may.exe
This version brings some changes to ABR/CBR, so results regarding those modes would be helpfull.
(because of internet access problems, I am directly providing the test version.)
Full Version: Lame test version - may 2005
Is this one compiled from the current cvs sources?
I would like to test it as well, but my OS dont support Win32 executables.
I would like to test it as well, but my OS dont support Win32 executables.
Try it using Wine.
WINE is Not an Emulator.
He's running a PowerBook G4.
He's running a PowerBook G4.
QUOTE
Is this one compiled from the current cvs sources
No. My home DSL connection is not restored, so I do not have cvs access.
QUOTE(Gabriel @ May 13 2005, 11:52 AM)
QUOTE
Is this one compiled from the current cvs sources
No. My home DSL connection is not restored, so I do not have cvs access.
If it's of any value to you, if you were to send me the amended source files, I could make them, and ICL compiles, available at Rarewares.
QUOTE(Althalus @ May 13 2005, 12:20 PM)
WINE is Not an Emulator.
He's running a PowerBook G4.
He's running a PowerBook G4.
My mistake. Sorry. Virtual PC then.
QUOTE(Latexxx @ May 13 2005, 02:21 PM)
QUOTE(Althalus @ May 13 2005, 12:20 PM)
WINE is Not an Emulator.
He's running a PowerBook G4.
He's running a PowerBook G4.
My mistake. Sorry. Virtual PC then.
$129 for MS Virtual PC + even more for a MS Windows license. I think not...
Then I rather compile lame from CVS. But that dont help Gabriel in this case.
QUOTE(krmathis @ May 13 2005, 06:18 PM)
QUOTE(Latexxx @ May 13 2005, 02:21 PM)
QUOTE(Althalus @ May 13 2005, 12:20 PM)
WINE is Not an Emulator.
He's running a PowerBook G4.
He's running a PowerBook G4.
My mistake. Sorry. Virtual PC then.
$129 for MS Virtual PC + even more for a MS Windows license. I think not...
Then I rather compile lame from CVS. But that dont help Gabriel in this case.
Well, you can get an 180-day test drive edition of Windows Server 2003 for free and at least the Windows version if Virtual PC is available as demo for 45 days.
QUOTE(john33 @ May 13 2005, 08:59 AM)
QUOTE(Gabriel @ May 13 2005, 11:52 AM)
QUOTE
Is this one compiled from the current cvs sources
No. My home DSL connection is not restored, so I do not have cvs access.
If it's of any value to you, if you were to send me the amended source files, I could make them, and ICL compiles, available at Rarewares.
I can also upload them to the CVS, if that's of any help.
QUOTE
I can also upload them to the CVS, if that's of any help.
Thanks, but no real need right now.
This is really a TEST version, as I am not sure about the positive impact of all the changes I made.
QUOTE(Gabriel @ May 14 2005, 03:49 AM)
QUOTE
I can also upload them to the CVS, if that's of any help.
Thanks, but no real need right now.
This is really a TEST version, as I am not sure about the positive impact of all the changes I made.
Can you suggest areas we might want to check against?
Yes, ABR/CBR (as mentionned in first post)
My apologies, I should have read it closer. Although It will be hard for me to test on my linux box without source.
I had a go abxing this with 396, but... my ears.. no results yet... Ill try lower bitrates soon.
To confirm there was some difference between, i transcoded a clip back and forth between 112kbs abr and cbr 20 times - not too helpful surely, but there is a difference.
Cleanish encode
396 transcode
97maytranscode
I guess thats enough of that, ill give 80kbs a proper go in day or two.
To confirm there was some difference between, i transcoded a clip back and forth between 112kbs abr and cbr 20 times - not too helpful surely, but there is a difference.
Cleanish encode
396 transcode
97maytranscode
I guess thats enough of that, ill give 80kbs a proper go in day or two.
Vaguely OT note: Mac or Linux users may want to snag a copy of QEMU, which is a free/opensource multiplatform emulator capable of booting up Windows and other OSs. No need to shell out for Virtual PC .
.
I've encoded a complete album for my portable player, and I was disappointed while listening to it. Back to home, and despite of my limited free time, I've decided to launch a quick listening test.
setting = --abr 130
samples = classical music (my usual sample suit)
tool : abchr for java, offset removed, gain, ~first second discarded. ABX between both encoded files.
RESULTS:
These results confirm that something wrong occurs with the new tunings. Sounds suffers from background alteration (noise is irregular, and musical informations are sometimes fluctuating too), a bit like 3.96.1. Slight progress noticed with the Bruhns (organ) sample, but that's all.
EDIT: abx logs are available here.
setting = --abr 130
samples = classical music (my usual sample suit)
tool : abchr for java, offset removed, gain, ~first second discarded. ABX between both encoded files.
RESULTS:
These results confirm that something wrong occurs with the new tunings. Sounds suffers from background alteration (noise is irregular, and musical informations are sometimes fluctuating too), a bit like 3.96.1. Slight progress noticed with the Bruhns (organ) sample, but that's all.
EDIT: abx logs are available here.
@guruboolez
Many thanks, guru. Glad to see you alive and well. I was hoping you might find a bit of time for this ... I'd love to help, myself, but these 63-year-old ears just aren't up to the task.
@Gabriel
Do you still need more testing of the May version, or are guru's results enough?
Regards,
Madrigal
Many thanks, guru. Glad to see you alive and well. I was hoping you might find a bit of time for this ... I'd love to help, myself, but these 63-year-old ears just aren't up to the task.
@Gabriel
Do you still need more testing of the May version, or are guru's results enough?
Regards,
Madrigal
QUOTE
Do you still need more testing of the May version, or are guru's results enough?
At least 1 CBR test would be nice.
i tested a10 against the may test version without a parameter (should be cbr128 jstereo) with the castenets sample and the first 2.5 sec of the first track of star wars ep1 ost.
i could easily abx the castanet samples against the originals but failed to abx the two mp3s.
with the starwars sample i experienced some kind of dropout or "blop" with the may test version. it was very hard for me to find this out as i listened to some sort of ringing percussion during wav vs mp3. i was able to abx the blop in the may-mp3 against the a10-mp3 with 6/6.
my conclusion is that with cbr128 at this sampe (which i choosed out of the blue
), i can confirm the results of guruboolez' test.
logs are here:
foo_abx v1.2 report
foobar2000 v0.8.3
2005/05/24 23:57:43
File A: file://E:\Musik\Klassik\Star Wars\1999 - Star Wars Episode I - The Phantom Menace.wv
File B: file://D:\Temp\tests\a10 vs may\ep1 01 a10.mp3
23:58:09 : Test started.
23:59:52 : 01/01 50.0%
00:01:37 : 02/02 25.0%
00:01:58 : 03/03 12.5%
00:02:06 : 04/04 6.3%
00:02:23 : 05/05 3.1%
00:02:30 : 06/06 1.6%
00:02:42 : 07/07 0.8%
00:02:50 : 08/08 0.4%
00:02:58 : 09/09 0.2%
00:03:22 : 09/10 1.1%
00:03:50 : 09/11 3.3%
00:04:09 : 10/12 1.9%
00:04:17 : 11/13 1.1%
00:04:25 : 12/14 0.6%
00:04:29 : Test finished.
----------
Total: 12/14 (0.6%)
foo_abx v1.2 report
foobar2000 v0.8.3
2005/05/25 00:04:43
File A: file://E:\Musik\Klassik\Star Wars\1999 - Star Wars Episode I - The Phantom Menace.wv
File B: file://D:\Temp\tests\a10 vs may\ep1 01 may.mp3
00:05:10 : Test started.
00:08:21 : 01/01 50.0%
00:08:31 : 01/02 75.0%
00:09:17 : 02/03 50.0%
00:09:26 : 03/04 31.3%
00:09:50 : 04/05 18.8%
00:10:07 : 05/06 10.9%
00:10:20 : 06/07 6.3%
00:10:32 : 07/08 3.5%
00:10:48 : 08/09 2.0%
00:11:00 : 09/10 1.1%
00:11:19 : 10/11 0.6%
00:11:38 : 11/12 0.3%
00:11:42 : Test finished.
----------
Total: 11/12 (0.3%)
foo_abx v1.2 report
foobar2000 v0.8.3
2005/05/25 00:12:03
File A: file://D:\Temp\tests\a10 vs may\ep1 01 a10.mp3
File B: file://D:\Temp\tests\a10 vs may\ep1 01 may.mp3
00:12:25 : Test started.
00:13:52 : 01/01 50.0%
00:14:25 : 02/02 25.0%
00:14:47 : 03/03 12.5%
00:17:15 : 04/04 6.3%
00:19:22 : 05/05 3.1%
00:20:31 : 06/06 1.6%
00:20:44 : Test finished.
----------
Total: 6/6 (1.6%)
i could easily abx the castanet samples against the originals but failed to abx the two mp3s.
with the starwars sample i experienced some kind of dropout or "blop" with the may test version. it was very hard for me to find this out as i listened to some sort of ringing percussion during wav vs mp3. i was able to abx the blop in the may-mp3 against the a10-mp3 with 6/6.
my conclusion is that with cbr128 at this sampe (which i choosed out of the blue
), i can confirm the results of guruboolez' test.logs are here:
CODE
foo_abx v1.2 report
foobar2000 v0.8.3
2005/05/24 23:57:43
File A: file://E:\Musik\Klassik\Star Wars\1999 - Star Wars Episode I - The Phantom Menace.wv
File B: file://D:\Temp\tests\a10 vs may\ep1 01 a10.mp3
23:58:09 : Test started.
23:59:52 : 01/01 50.0%
00:01:37 : 02/02 25.0%
00:01:58 : 03/03 12.5%
00:02:06 : 04/04 6.3%
00:02:23 : 05/05 3.1%
00:02:30 : 06/06 1.6%
00:02:42 : 07/07 0.8%
00:02:50 : 08/08 0.4%
00:02:58 : 09/09 0.2%
00:03:22 : 09/10 1.1%
00:03:50 : 09/11 3.3%
00:04:09 : 10/12 1.9%
00:04:17 : 11/13 1.1%
00:04:25 : 12/14 0.6%
00:04:29 : Test finished.
----------
Total: 12/14 (0.6%)
foo_abx v1.2 report
foobar2000 v0.8.3
2005/05/25 00:04:43
File A: file://E:\Musik\Klassik\Star Wars\1999 - Star Wars Episode I - The Phantom Menace.wv
File B: file://D:\Temp\tests\a10 vs may\ep1 01 may.mp3
00:05:10 : Test started.
00:08:21 : 01/01 50.0%
00:08:31 : 01/02 75.0%
00:09:17 : 02/03 50.0%
00:09:26 : 03/04 31.3%
00:09:50 : 04/05 18.8%
00:10:07 : 05/06 10.9%
00:10:20 : 06/07 6.3%
00:10:32 : 07/08 3.5%
00:10:48 : 08/09 2.0%
00:11:00 : 09/10 1.1%
00:11:19 : 10/11 0.6%
00:11:38 : 11/12 0.3%
00:11:42 : Test finished.
----------
Total: 11/12 (0.3%)
foo_abx v1.2 report
foobar2000 v0.8.3
2005/05/25 00:12:03
File A: file://D:\Temp\tests\a10 vs may\ep1 01 a10.mp3
File B: file://D:\Temp\tests\a10 vs may\ep1 01 may.mp3
00:12:25 : Test started.
00:13:52 : 01/01 50.0%
00:14:25 : 02/02 25.0%
00:14:47 : 03/03 12.5%
00:17:15 : 04/04 6.3%
00:19:22 : 05/05 3.1%
00:20:31 : 06/06 1.6%
00:20:44 : Test finished.
----------
Total: 6/6 (1.6%)
This sample is pretty artificial, as there is no channel correlation. Still, thought you might find it interesting. Original and encoded versions included.
http://www.hydrogenaudio.org/forums/index....showtopic=34329
To my ears at preset cbr 128, 3.90.3 performs substantially better with this sample than 3.96.1/3.97a8. The later versions sound pretty horrible by comparison.
I can reliably abx the sample between 3.96.1 and lame-may, with lame-may very slightly better (more tonality in the "bells")
I cannot abx the sample between 3.97a8 and lame-may at all.
http://www.hydrogenaudio.org/forums/index....showtopic=34329
To my ears at preset cbr 128, 3.90.3 performs substantially better with this sample than 3.96.1/3.97a8. The later versions sound pretty horrible by comparison.
I can reliably abx the sample between 3.96.1 and lame-may, with lame-may very slightly better (more tonality in the "bells")
I cannot abx the sample between 3.97a8 and lame-may at all.
QUOTE(Linux Zealot Troll @ May 27 2005, 12:50 PM)
To my ears at preset cbr 128, 3.90.3 performs substantially better with this sample than 3.96.1/3.97a8. The later versions sound pretty horrible by comparison.
I can reliably abx the sample between 3.96.1 and lame-may, with lame-may very slightly better (more tonality in the "bells")
I cannot abx the sample between 3.97a8 and lame-may at all.
Surely the only meaningful comparison, at this point, is between 3.97a10 and may 2005, as guruboolez and ToS_Maverick have done.I can reliably abx the sample between 3.96.1 and lame-may, with lame-may very slightly better (more tonality in the "bells")
I cannot abx the sample between 3.97a8 and lame-may at all.
Regards,
Madrigal
QUOTE(Madrigal @ May 27 2005, 08:33 PM)
QUOTE(Linux Zealot Troll @ May 27 2005, 12:50 PM)
To my ears at preset cbr 128, 3.90.3 performs substantially better with this sample than 3.96.1/3.97a8. The later versions sound pretty horrible by comparison.
I can reliably abx the sample between 3.96.1 and lame-may, with lame-may very slightly better (more tonality in the "bells")
I cannot abx the sample between 3.97a8 and lame-may at all.
Surely the only meaningful comparison, at this point, is between 3.97a10 and may 2005, as guruboolez and ToS_Maverick have done.I can reliably abx the sample between 3.96.1 and lame-may, with lame-may very slightly better (more tonality in the "bells")
I cannot abx the sample between 3.97a8 and lame-may at all.
Regards,
Madrigal
Sorry, thought I was using the latest alpha.
I do not consider lame-may to be good as alpha-10 with this sample, the "bells" have more noise.
foo_abx v1.2 report
foobar2000 v0.8.3
2005/05/27 20:58:00
File A: file://C:\Documents and Settings\M\My Documents\es50-397a10.mp3
File B: file://C:\Documents and Settings\M\My Documents\es50-may.mp3
20:58:00 : Test started.
20:58:51 : 01/01 50.0%
20:59:10 : 02/02 25.0%
20:59:42 : 03/03 12.5%
21:00:05 : 04/04 6.3%
21:00:53 : 05/05 3.1%
21:01:43 : 06/06 1.6%
21:02:17 : 07/07 0.8%
21:02:23 : Test finished.
----------
Total: 7/7 (0.8%)
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.