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: A personal LAME APS problem tracks test. (Read 19920 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

A personal LAME APS problem tracks test.

Lately i keep finding troubled tracks on my music collection (and still finding  ) that are encoded by LAME 3.97 -V 2 --vbr-new, for some reason i am picking up artifacts more easy and i have not upgraded sound hardware and did not do any artifact training. Am shocked that i managed to ABX those tracks with a cheap Sound Blaster Audigy SE sound card and my Sennheiser HD 202 headphones.

Anthrax - Among The Living

LAME 3.97 -V 2 --vbr-new

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.4.4
2007/10/24 00:36:00

File A: F:\Listen Tests\Among The Living.flac
File B: E:\Music\Albums\Anthrax - Among The Living\01. Anthrax - Among The Living.mp3

00:36:00 : Test started.
00:36:32 : 01/01  50.0%
00:36:59 : 02/02  25.0%
00:37:27 : 03/03  12.5%
00:38:12 : 04/04  6.3%
00:38:25 : 05/05  3.1%
00:38:36 : 06/06  1.6%
00:38:45 : 07/07  0.8%
00:38:58 : 08/08  0.4%
00:39:10 : 09/09  0.2%
00:39:23 : 10/10  0.1%
00:39:25 : Test finished.

----------
Total: 10/10 (0.1%)


At 2:10 there is a sharp precho sound, and i found this track is transparent on 3.90.3 and 3.98b6.

Metallica - Harvester Of Sorrow

LAME 3.90.3 APS

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5 beta 5
2007/11/26 23:09:58

File A: F:\Listen Tests\Harvester Of Sorrow.wav.mp3
File B: F:\Listen Tests\Harvester Of Sorrow.wav

23:09:58 : Test started.
23:10:17 : 00/01  100.0%
23:10:27 : 01/02  75.0%
23:10:38 : 02/03  50.0%
23:11:01 : 03/04  31.3%
23:11:21 : 04/05  18.8%
23:11:59 : 05/06  10.9%
23:12:11 : 06/07  6.3%
23:12:48 : 07/08  3.5%
23:13:00 : 08/09  2.0%
23:13:11 : 09/10  1.1%
23:13:33 : 10/11  0.6%
23:14:16 : 11/12  0.3%
23:14:17 : Test finished.

----------
Total: 11/12 (0.3%)


A preecho at 0:04.

LAME 3.97 -V 2 --vbr-new

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5 beta 5
2007/11/26 22:42:17

File A: F:\Listen Tests\Harvester Of Sorrow.flac
File B: E:\Music\Albums\Metallica - ...And Justice For All\06. Metallica - Harvester Of Sorrow.mp3

22:42:17 : Test started.
22:42:38 : 01/01  50.0%
22:42:57 : 02/02  25.0%
22:43:25 : 03/03  12.5%
22:43:50 : 04/04  6.3%
22:44:44 : 05/05  3.1%
22:45:13 : 06/06  1.6%
22:45:37 : 07/07  0.8%
22:46:10 : 08/08  0.4%
22:46:51 : 09/09  0.2%
22:47:51 : 10/10  0.1%
22:47:53 : Test finished.

----------
Total: 10/10 (0.1%)


A preecho at 0:04, but sounds worse then the 3.90.3 encode.

LAME 3.98b6 (Nov 1) -V 2

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5 beta 5
2007/11/26 23:26:05

File A: F:\Listen Tests\Harvester Of Sorrow LAME3.98b6 V2.mp3
File B: F:\Listen Tests\Harvester Of Sorrow.flac

23:26:05 : Test started.
23:26:35 : 01/01  50.0%
23:27:06 : 02/02  25.0%
23:27:40 : 02/03  50.0%
23:27:54 : 02/04  68.8%
23:28:20 : 03/05  50.0%
23:28:26 : 04/06  34.4%
23:28:36 : 05/07  22.7%
23:29:10 : 06/08  14.5%
23:29:30 : 06/09  25.4%
23:29:51 : 07/10  17.2%
23:30:19 : 08/11  11.3%
23:30:44 : 09/12  7.3%
23:30:52 : Test finished.

----------
Total: 9/12 (7.3%)


Am not sure with this encode but its better then the last 2 encodes.

Ministry - Thieves

LAME 3.90.3 APS

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5 beta 5
2007/11/26 21:28:48

File A: F:\Listen Tests\Thieves.wav
File B: F:\Listen Tests\Thieves.wav.mp3

21:28:48 : Test started.
21:29:13 : 01/01  50.0%
21:29:29 : 02/02  25.0%
21:30:04 : 03/03  12.5%
21:30:23 : 04/04  6.3%
21:31:01 : 05/05  3.1%
21:31:34 : 06/06  1.6%
21:32:16 : 07/07  0.8%
21:32:42 : 08/08  0.4%
21:32:58 : 09/09  0.2%
21:33:22 : 10/10  0.1%
21:33:25 : Test finished.

----------
Total: 10/10 (0.1%)


A weird popping like noise at the start of the track.

LAME 3.98b6 -V 2

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5 beta 5
2007/11/26 21:43:57

File A: F:\Listen Tests\Thieves V2 LAME3.98b6.mp3
File B: F:\Listen Tests\Thieves.wav

21:43:57 : Test started.
21:44:45 : 01/01  50.0%
21:45:16 : 02/02  25.0%
21:46:26 : 02/03  50.0%
21:46:49 : 03/04  31.3%
21:47:16 : 04/05  18.8%
21:47:45 : 05/06  10.9%
21:47:59 : 05/07  22.7%
21:49:41 : 06/08  14.5%
21:49:58 : 07/09  9.0%
21:52:06 : 07/10  17.2%
21:52:17 : 07/11  27.4%
21:52:20 : Test finished.

----------
Total: 7/11 (27.4%)


Am not sure with this encode i keep thinking there is something wrong with the cymbals in backgroud.

At the moment am finding LAME 3.98b to be better then LAME 3.97 at V2 and better or good as LAME 3.90.3 --alt-preset like on my early findings a few months ago.
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #1
Which compile of 3.98b6 are you using for your test?

EDIT: I just noticed the Nov 1 in your post... you should check RareWares, I think it was updated today.

A personal LAME APS problem tracks test.

Reply #2
You could try -V1 then --abr 260.. If you can still abx here and there, you might just have to accept that mp3 is not fully transparent to you.

A personal LAME APS problem tracks test.

Reply #3
Which compile of 3.98b6 are you using for your test?

EDIT: I just noticed the Nov 1 in your post... you should check RareWares, I think it was updated today.


I used the Nov 1 build, but beta 6 is not offical build its a CVS fork of updates and it some has a major bug fix from beta 5.

You could try -V1 then --abr 260.. If you can still abx here and there, you might just have to accept that mp3 is not fully transparent to you.


I tried -V 1 --vbr-new on LAME 3.97 on some tracks i found a few months ago that was transparent or just sounded a little better and V0 did not make a difference at all over V1. And ATM am finding V0 to be useless and a waste of space since it can be possible that its not transparent on tracks that struggle at V2 and just sounds a little better and gain 10mb - 20mb on a album at V0, and i keep getting bitrates at 280 on V0 so i might use V1 instead or just stick to V2.
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #4
a) With pre-echo we have to be content with a rather modest behavior when using mp3 I'm afraid. But as you've found out there is still progress with Lame, and as you're sensitive towards pre-echo you may have your part in further improving Lame.
b) From former listening tests I also experienced that going from -V2 to -V1 can improve certain problematic samples in a meaningful way, but going from -V1 to -V0 isn't worth the price. But you may consider trying --abr 260 or similar as shadowking suggested: if you allow for very high bitrate this is the best way IMO to get at the best quality possible. This experience of mine is based on non-pre-echo problems and 3.97. 3.98beta's psy model/VBR method has improved a lot since, so this may not be valid any more. Anyway I would give --abr 260 or similar a try.
lame3995o -Q1.7 --lowpass 17

A personal LAME APS problem tracks test.

Reply #5
a) With pre-echo we have to be content with a rather modest behavior when using mp3 I'm afraid. But as you've found out there is still progress with Lame, and as you're sensitive towards pre-echo you may have your part in further improving Lame.
b) From former listening tests I also experienced that going from -V2 to -V1 can improve certain problematic samples in a meaningful way, but going from -V1 to -V0 isn't worth the price. But you may consider trying --abr 260 or similar as shadowking suggested: if you allow for very high bitrate this is the best way IMO to get at the best quality possible. This experience of mine is based on non-pre-echo problems and 3.97. 3.98beta's psy model/VBR method has improved a lot since, so this may not be valid any more. Anyway I would give --abr 260 or similar a try.

I could give ABR 260 a try or 230, but i did try ABR 128 on 3.97 a couple years ago but i found it sounded the same or just a tad better then 128 CBR but that mode was using very low bitrates anyway, but ABR could out-perform VBR if the ABR target is over the VBR mode target since ABR target will be fixed.
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #6
The main difference between VBR and ABR is to me: the bitrate control of VBR relies very much on the psy model whereas the bitrate control of ABR doesn't, and bitrate variation of ABR is more conservative as a tendency. At very high bitrate like 260 kbps the advantages of ABR are more important IMO than the disadvantages: bitrate is very high at any rate to ensure transparency in most cases, and those rare situations where the psy models flaws doesn't have the same bad impact it has when using VBR.

So the situation with very high bitrate is quite favorable for using ABR - quite a different thing then when using low bitrate.

Whether or not this improves things for your pre-echo problems is another story. As a tendency VBR usually does a good job here. Anyway I would give ABR a try.
lame3995o -Q1.7 --lowpass 17

A personal LAME APS problem tracks test.

Reply #7
Your reasoning makes no sense because even in ABR mode the psymodel still controls the bit allocation, and just 260kbps, let alone 260kbps average, is not enough to guarantee transparency in critical situations.

If the psymodel is truly off, the instantenous bitrate at that frame should be (very roughly) >368kbps to ensure no artifacts can happen [1].

ABR will not protect you from that and will constrain the encoder when it KNOWS it must go up to the maximal bitrate.

So think: what is the more typical case, the psymodel is right or the psymodel is wrong?

[1] Gives about 18dB SNR over the entire frequency range.

A personal LAME APS problem tracks test.

Reply #8
The logic is simple: The higher you go with ABR bitrate the better gets the chance/risk relation of using ABR vs. VBR simply because you can't get better than using 320 kbps frames, and because of the fact that VBR introduces a certain risk. That's why you can read of ABR in the Lame documentation as being the 'safe' vbr method. BTW due to restrictions of some encoders concerning bit reservoir usage of 320 kbps frames it can even happen that you get at a better net audio data rate when using 256 kbps frames at a critical spot.

Anyway that's theory. I do have some experience with problem samples behaving better with ABR at ~260 kbps, but these weren't pre-echo samples. For pre-echo issues I doubt ABR will improve things but would give it a try.
lame3995o -Q1.7 --lowpass 17

A personal LAME APS problem tracks test.

Reply #9
--

A personal LAME APS problem tracks test.

Reply #10
mnt,

You might give the 10 Dec. 07 b6 compile on Rarewares a try to see if it has gotten any better than b5.
Zune 80, Tak -p4 audio library, Lossless=Choice

A personal LAME APS problem tracks test.

Reply #11
I gave up on LAME V2 because on my home stereo system, which is a fairly decent (not very expensive though) system... I can't take MP3's anymore... They pretty have a little less of bass frequency. The original CD's have got just a bit more bass and punch.
Care to put some objective proof behind these flippant and ill-advised remarks?  Need I remind you of TOS #8?
I'm on a horse.

A personal LAME APS problem tracks test.

Reply #12
I gave up on LAME V2 because on my home stereo system, which is a fairly decent (not very expensive though) system... I can't take MP3's anymore... They pretty have a little less of bass frequency. The original CD's have got just a bit more bass and punch.


It could be bad decoder or equilisation, hardware fault. I suggest getting some dedicated mp3 device for reproduction.

Ivan.
Error 404; signature server not available.

A personal LAME APS problem tracks test.

Reply #13
mnt,

You might give the 10 Dec. 07 b6 compile on Rarewares a try to see if it has gotten any better than b5.


I have done a quick test on the new build, also now LAME 3.98 now has a new psy model that is a based on nspsytune and is on default on vbr-new so now lame 3.98 needs more listening tests.

Stigmata -V 2 (LAME3.98b6 Dec 10)
Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5 beta 7
2007/12/11 15:04:22

File A: F:\Listen Tests\Stigmata.wav
File B: F:\Listen Tests\Stigmata (LAME 3.98b6 Dec10 V2.mp3

15:04:22 : Test started.
15:05:00 : 01/01  50.0%
15:05:32 : 02/02  25.0%
15:05:50 : 03/03  12.5%
15:06:01 : 04/04  6.3%
15:06:25 : 05/05  3.1%
15:07:18 : 06/06  1.6%
15:07:40 : 07/07  0.8%
15:07:54 : 08/08  0.4%
15:08:31 : 09/09  0.2%
15:08:54 : 10/10  0.1%
15:09:04 : Test finished.

----------
Total: 10/10 (0.1%)


This encode sounds then the other versions of LAME i tested, or i got more sensitive towards pre-echo  but on LAME 3.97 i find this track to be transparent at V0 -vbr-new on but can ABX it easy on V2 and V1 -vbr-new.
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #14
I found yet another precho artifact on track when i was trying out -V 1 --vbr-new on 3.97, which is transparent on 3.90.3 APS. The track is Back The F**k Up by Fear Factory from the album Digimortal, also I do have another track from them (Replica) that is not transparent on V2, but not tested that on higher setting atm.

LAME 3.97 -V 2 --vbr-new

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1 beta 1
2008/02/10 13:14:34

File A: E:\Temp\Listen Tests\Back The F**k Up V2.mp3
File B: E:\Temp\Listen Tests\Back The F**k Up.flac

13:14:34 : Test started.
13:16:05 : 01/01  50.0%
13:16:11 : 02/02  25.0%
13:16:20 : 03/03  12.5%
13:16:48 : 04/04  6.3%
13:17:08 : 05/05  3.1%
13:17:21 : 06/06  1.6%
13:17:29 : 07/07  0.8%
13:17:41 : 08/08  0.4%
13:18:51 : 09/09  0.2%
13:19:33 : 10/10  0.1%
13:19:35 : Test finished.

----------
Total: 10/10 (0.1%)


Precho at 0.29

LAME 3.97 -V 1 --vbr-new

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1 beta 1
2008/02/10 13:04:33

File A: E:\Temp\Listen Tests\Back The F**k Up.mp3
File B: E:\Temp\Listen Tests\Back The F**k Up.flac

13:04:33 : Test started.
13:05:51 : 01/01  50.0%
13:06:13 : 02/02  25.0%
13:06:31 : 03/03  12.5%
13:07:22 : 04/04  6.3%
13:07:35 : 05/05  3.1%
13:07:53 : 06/06  1.6%
13:08:10 : 07/07  0.8%
13:08:31 : 08/08  0.4%
13:08:49 : 09/09  0.2%
13:09:26 : 10/10  0.1%
13:09:27 : Test finished.

----------
Total: 10/10 (0.1%)


Same artifact, possibly worse then the V2 encode.

LAME 3.97 -V 0 --vbr-new

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1 beta 1
2008/02/10 18:01:47

File A: E:\Temp\Listen Tests\Back The F**k Up V0.mp3
File B: E:\Temp\Listen Tests\Back The F**k Up.flac

18:01:47 : Test started.
18:02:50 : 01/01  50.0%
18:03:30 : 02/02  25.0%
18:04:19 : 03/03  12.5%
18:04:29 : 04/04  6.3%
18:04:51 : 05/05  3.1%
18:05:47 : 06/06  1.6%
18:06:16 : 06/07  6.3%
18:06:40 : 07/08  3.5%
18:07:37 : 08/09  2.0%
18:08:12 : 09/10  1.1%
18:08:39 : 10/11  0.6%
18:09:43 : 11/12  0.3%
18:09:47 : Test finished.

----------
Total: 11/12 (0.3%)


The artifact is still there but its only a tad better.

LAME 3.98 Beta 6 -V 2

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1 beta 1
2008/02/10 18:16:05

File A: E:\Temp\Listen Tests\Back The F**k Up.flac
File B: E:\Temp\Listen Tests\Back The F**k Up LAME3.98b6 V2.mp3

18:16:05 : Test started.
18:18:05 : 00/01  100.0%
18:18:16 : 01/02  75.0%
18:18:33 : 02/03  50.0%
18:18:50 : 03/04  31.3%
18:19:48 : 04/05  18.8%
18:20:22 : 04/06  34.4%
18:20:44 : 05/07  22.7%
18:21:06 : 05/08  36.3%
18:21:23 : 06/09  25.4%
18:21:48 : 07/10  17.2%
18:22:02 : 08/11  11.3%
18:22:24 : 09/12  7.3%
18:22:33 : 10/13  4.6%
18:22:46 : 11/14  2.9%
18:23:01 : 12/15  1.8%
18:23:19 : 13/16  1.1%
18:23:30 : 14/17  0.6%
18:23:42 : 15/18  0.4%
18:23:57 : 16/19  0.2%
18:24:36 : 17/20  0.1%
18:24:39 : Test finished.

----------
Total: 17/20 (0.1%)


Very strange, no precho artifact this time. For reason at around 0:29 - 0:33 seems less audible and lifeless on this encode, like some sounds have been removed.

After playing with V1 and V0, I found that V0 seems to be really pointless if you can ABX V2 and only sounds a tad better or sometimes transparent at V0.
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #15
I have tried the other track that I mentioned on my last post.

LAME 3.97 -V 2 --vbr-new

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1 beta 1
2008/02/11 21:12:34

File A: E:\Temp\Listen Tests\Replica.wav
File B: E:\Music\Albums\Fear Factory - Demanufacture\04. Fear Factory - Replica.mp3

21:12:34 : Test started.
21:13:14 : 01/01  50.0%
21:13:31 : 02/02  25.0%
21:14:25 : 03/03  12.5%
21:14:47 : 04/04  6.3%
21:15:06 : 05/05  3.1%
21:15:28 : 06/06  1.6%
21:15:34 : 07/07  0.8%
21:15:53 : 08/08  0.4%
21:16:51 : 09/09  0.2%
21:17:02 : 10/10  0.1%
21:17:55 : 11/11  0.0%
21:18:30 : 12/12  0.0%
21:19:57 : Test finished.

----------
Total: 12/12 (0.0%)


I was gonna ABX the precho the artifact at 0:12, but I just found and ABXed this psst pop (precho) which which appears alot on the eig sample, but it is hard to notice if you are not concentrateing on the track.

I tested on 3.98 Beta 6 at around 0:12 and got 18/23, sometimes I could hear the artifact. But my hearing has gotten abit bad or back to normal every since I had flu at the Christmas holidays, I now have trouble ABX some the tracks I did a few months ago.
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #16
-------

A personal LAME APS problem tracks test.

Reply #17
this is all heavy metal, right? it may just be the particular kind of music not to be very friendly to mp3.

Well most of the time I keep spotting artifacts on drums on metal,I get the impression that LAME really hates Industrial Metal.
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #18
I found a track this weekend that vbr-old does better then vbr-new on LAME 3.97, which is not transparent on 192cbr - 224cbr (i have tried 256 though), but its transparent on -V 0 --vbr-new.

LAME 3.97 192kbps CBR

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1
2008/03/23 15:25:24

File A: E:\Temp\Listen Tests\Operation Ground And Pound 192cbr.mp3
File B: F:\Music\Rips\DragonForce - Inhuman Rampage\04. Operation Ground And Pound.flac

15:25:24 : Test started.
15:25:49 : 01/01  50.0%
15:26:13 : 02/02  25.0%
15:26:38 : 03/03  12.5%
15:26:46 : 04/04  6.3%
15:27:39 : 05/05  3.1%
15:28:31 : 06/06  1.6%
15:28:40 : 07/07  0.8%
15:29:13 : 08/08  0.4%
15:29:30 : 09/09  0.2%
15:29:43 : 10/10  0.1%
15:29:44 : Test finished.

----------
Total: 10/10 (0.1%)


Warbling and precho at the start of the track.

LAME 3.97 224kpbs CBR

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1
2008/03/23 15:40:10

File A: F:\Music\Rips\DragonForce - Inhuman Rampage\04. Operation Ground And Pound.flac
File B: E:\Temp\Listen Tests\Operation Ground And Pound 224cbr.mp3

15:40:10 : Test started.
15:40:30 : 01/01  50.0%
15:41:11 : 02/02  25.0%
15:41:53 : 03/03  12.5%
15:43:03 : 04/04  6.3%
15:44:12 : 05/05  3.1%
15:45:27 : 06/06  1.6%
15:45:46 : 07/07  0.8%
15:46:35 : 08/08  0.4%
15:48:50 : 09/09  0.2%
15:50:54 : 10/10  0.1%
15:50:56 : Test finished.

----------
Total: 10/10 (0.1%)


A precho at the start. Sounds alot better the 192kbps and almost transparent.

LAME 3.97 -V 2 --vbr-new

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1
2008/03/23 12:30:40

File A: F:\Music\Rips\DragonForce - Inhuman Rampage\04. Operation Ground And Pound.flac
File B: C:\Music\Albums\DragonForce - Inhuman Rampage\04. Operation Ground And Pound.mp3

12:30:40 : Test started.
12:31:12 : 01/01  50.0%
12:31:31 : 02/02  25.0%
12:31:44 : 03/03  12.5%
12:32:08 : 04/04  6.3%
12:33:04 : 05/05  3.1%
12:33:47 : 06/06  1.6%
12:34:26 : 07/07  0.8%
12:35:21 : 08/08  0.4%
12:35:54 : 09/09  0.2%
12:36:16 : 10/10  0.1%
12:36:24 : Test finished.

----------
Total: 10/10 (0.1%)


A precho at the start sounds better then the 192kbps encode, but worse then the 224 encode.

LAME 3.97 -V 2 --vbr-old

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1
2008/03/23 15:54:30

File A: F:\Music\Rips\DragonForce - Inhuman Rampage\04. Operation Ground And Pound.flac
File B: E:\Temp\Listen Tests\Operation Ground And Pound V2 vbr-old.mp3

15:54:30 : Test started.
15:57:03 : 00/01  100.0%
15:57:48 : 01/02  75.0%
15:57:57 : 02/03  50.0%
15:58:30 : 03/04  31.3%
15:58:44 : 04/05  18.8%
15:59:20 : 05/06  10.9%
16:00:09 : 05/07  22.7%
16:01:48 : 05/08  36.3%
16:02:45 : 06/09  25.4%
16:03:53 : 07/10  17.2%
16:04:08 : Test finished.

----------
Total: 7/10 (17.2%)


Am not really sure about this encode.

LAME 3.98 Beta 6 (12 March 2008)

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.5.1
2008/03/23 16:35:59

File A: F:\Music\Rips\DragonForce - Inhuman Rampage\04. Operation Ground And Pound.flac
File B: E:\Temp\Listen Tests\Operation Ground And Pound 3.98b6 V2.mp3

16:35:59 : Test started.
16:36:14 : 00/01  100.0%
16:36:19 : 01/02  75.0%
16:36:24 : 02/03  50.0%
16:36:43 : 03/04  31.3%
16:37:24 : 04/05  18.8%
16:37:40 : 05/06  10.9%
16:38:20 : 06/07  6.3%
16:38:57 : 07/08  3.5%
16:39:48 : 08/09  2.0%
16:39:56 : 09/10  1.1%
16:40:36 : 10/11  0.6%
16:40:38 : Test finished.

----------
Total: 10/11 (0.6%)


Sounds better then 3.97 possibly sounds the same as the 224 cbr encode, but -vbr-old (LAME 3.97) sounded better.

A link to a 30 sec sample of this track:

http://www.sendspace.com/file/pa2n74
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #19
---------

A personal LAME APS problem tracks test.

Reply #20
/mnt does this go away at 320 CBR 3.97 final?

I had a quick test at 256kbps CBR by using --preset cbr 256 and got 7/12, I struggled to ABX with this sample and also its getting late when i was testing it. I think it should go away at 320kbps, but like most of artifacts expect Stigmata by Ministry I picked up on my music collection goes away when I crank up the voulume which disorts the artifacts away such as Harvester Of Sorrow.
"I never thought I'd see this much candy in one mission!"

A personal LAME APS problem tracks test.

Reply #21
-----------

A personal LAME APS problem tracks test.

Reply #22
/mnt does this go away at 320 CBR 3.97 final?


Just did a test at preset insane (320kbps cbr) and got 7/12, I struggled with this test aswell I think it is transparent, but sometimes I think I might have heard the artifact. Anyway this sample shows that V2 is better then 192kbps CBR (it should anyway) and going from 192kbps - 224kpbs CBR can make a difference.
"I never thought I'd see this much candy in one mission!"

 

A personal LAME APS problem tracks test.

Reply #23
Any conclusions yet?

Could you somehow say that there's less artifact with V2 than wth V0 

A personal LAME APS problem tracks test.

Reply #24
Any conclusions yet?

Could you somehow say that there's less artifact with V2 than wth V0 

If that means me: No, according to my experience -V0 is better than -V2 at least on some occasion.
lame3995o -Q1.7 --lowpass 17