Lame 3.98b5-killer MP3 sample

2007-09-20 00:44:02

I know this file has been laying around, and it was already ABX'ed.
Could someone explain technically why Lame fails miserably with this particular sample?
Which category this sample fits? Rare? Occasional?
For those who are into more synth/electronic music, are these samples more frequent?

This sample also kills Nero AAC ~256vbr.
MPC is also killed at -q5/-q6/-q7/-q8. (Although less than Lame)
I am unable to ABX OGG over -q7 (~224) and it's almost good enough at -q6 (~192)

ABX logs:

LAME 3.98 beta 5 - using -V 0
----------------------------------
foo_abx 1.3.1 report
foobar2000 v0.9.4.3
2007/09/19 19:31:24

File A: D:\data\download\(V0)-3.98-eig.mp3
File B: D:\data\download\eig.wv

19:31:24 : Test started.
19:31:43 : 01/01  50.0%
19:31:59 : 02/02  25.0%
19:32:14 : 03/03  12.5%
19:32:38 : 04/04  6.3%
19:33:11 : 05/05  3.1%
19:34:06 : 06/06  1.6%
19:34:20 : 07/07  0.8%
19:34:30 : 08/08  0.4%
19:35:17 : 09/09  0.2%
19:35:40 : 10/10  0.1%
19:36:01 : 11/11  0.0%
19:36:20 : 12/12  0.0%
19:37:47 : 13/13  0.0%
19:38:02 : 14/14  0.0%
19:38:22 : 15/15  0.0%
19:40:36 : 16/16  0.0%
19:40:41 : Test finished.

----------
Total: 16/16 (0.0%)

-------------------------------------------------------------------------

LAME 3.98 beta 5 - using -b 320

foo_abx 1.3.1 report
foobar2000 v0.9.4.3
2007/09/19 20:07:23

File A: D:\data\download\eig.wv
File B: D:\data\download\(320)-3.98-eig.mp3

20:07:23 : Test started.
20:07:59 : 01/01  50.0%
20:08:10 : 02/02  25.0%
20:08:35 : 02/03  50.0%
20:08:59 : 03/04  31.3%
20:09:28 : 04/05  18.8%
20:09:46 : 05/06  10.9%
20:09:56 : 06/07  6.3%
20:10:21 : 07/08  3.5%
20:10:31 : 08/09  2.0%
20:10:43 : 09/10  1.1%
20:11:07 : 10/11  0.6%
20:11:21 : 11/12  0.3%
20:11:33 : 12/13  0.2%
20:11:42 : 12/14  0.6%
20:12:04 : 12/15  1.8%
20:12:24 : 13/16  1.1%
20:12:46 : 14/17  0.6%
20:13:01 : 15/18  0.4%
20:13:28 : 16/19  0.2%
20:13:48 : 17/20  0.1%
20:14:01 : 18/21  0.1%
20:14:14 : 19/22  0.0%
20:14:31 : 20/23  0.0%
20:15:01 : 21/24  0.0%
20:15:20 : 22/25  0.0%
20:15:33 : 23/26  0.0%
20:15:49 : 24/27  0.0%
20:16:07 : 25/28  0.0%
20:16:19 : 25/29  0.0%
20:16:38 : 26/30  0.0%
20:16:58 : 27/31  0.0%
20:17:35 : 28/32  0.0%
20:17:57 : 29/33  0.0%
20:18:09 : 30/34  0.0%
20:18:11 : Test finished.

----------

Here are the samples:
http://www.hydrogenaudio.org/forums/index....showtopic=57631

Lame 3.98b5-killer MP3 sample

Reply #1 – 2007-09-20 01:53:11

EIG is a strong pre echo impulse case. Rare in 'normal' music but can happen in artificial music. 3.98 is better than previous version though.

Lame 3.98b5-killer MP3 sample

Reply #2 – 2007-09-20 13:35:44

10 minutes to ABX? I wouldn't rate that "easy" or "killer" (in the sense of showstopper) at all, then.

I used to have a sample, computer generated music, that also had added noise due to pre-echo, ( awe32_30sec i think it was the name).
It was easy to spot at 320kbps with 3.96 (or was 3.97 alpha already? ), and is a thing that can't be completely fixed with mp3.  (the same sample with MPC at *-q5* uses 440kbps). Newer versions improved it, and therefore, most of the pre-echo samples, but probably this is as much as it can get.

Lame 3.98b5-killer MP3 sample

Reply #3 – 2007-09-22 00:26:13

How about MP3 freeformat @ 640 kbps?

Lame 3.98b5-killer MP3 sample

Reply #4 – 2007-09-22 22:07:55

Which player can handle that?
I've tested a few and nothing can play MP3 with freeformat

Lame 3.98b5-killer MP3 sample

Reply #5 – 2007-09-23 00:59:49

Wabbit jests.  From LAME documentation:

... even if an mp3 decoder is required to support free bitrates at least up to 320 kbps, many players are unable to deal with it.  Tests have shown that the following decoders support free format:

FreeAmp up to 440 kbps
in_mpg123 up to 560 kbps
l3dec up to 310 kbps
LAME up to 560 kbps
MAD up to 640 kbps

Lame 3.98b5-killer MP3 sample

Reply #6 – 2007-09-23 19:20:08

mdefranc....

I meant PLAYING the file.
Do you play these free-format files where?
Tell me...

Lame 3.98b5-killer MP3 sample

Reply #7 – 2007-09-24 02:12:22

I don't use free format.  High bit rates with the same algorithm won't necessarily reduce artifacts..  But, if you want playback of such an experiment, then try FreeAmp Player, CoolPlayer, MAD plugin for WinAmp2 Player, or in_mpg123 plugin for WinAmp5 Player.