High Bitrate Tests Options

[Seed]

The short sample in question is from "V-Tracks - Subway 26".

SK1 (a former member) and I detected a sample from a trance track that has
some pre-echo problems with LAME 3.90.3, even at higher settings like
--alt-preset insane.

I sat down and tested the sample, and indeed, 9/9, alhough I needed more than
an hour to finish this test. We decided to see if it's a codec-related problem
and prepared samples with MPC --insane, Vorbis GT3b2 -q 8, Nero AAC 6.3.0.3
at the VBR settings of Extreme and Audiophile, and finally, OptimFROG DualStream.

I concentrated on position 00:00.4 - 00:00.8. Throughout the sample, there are
repeating hi-hats on the left channel. They are smeared by the pre-echo effect.

In order to rule out any possible distortion/aliasing problems because of the
sound cards resampling from 44.1 KHz to 48 KHz, we used Naoki Shibata's SSRC
1.29 with these settings:

SSRC.exe --twopass --dither 0 --rate 48000 V-Tracks.wav V-Tracks-48.wav

OK, here we go:

I had mixed luck with Nero Extreme. I could not concentrate enough to get 5 in
a row, but I managed 4/4 about 5 times during 5 minutes. Unfortunately
WinABX froze, and this log can only shows my miserable attempts to
succeed again after running it for the second time. It's 4 AM and I will
probably quit now:

WinABX v0.41 test report
02/06/2004 00:51:41

A file: E:\V-Tracks-48.wav
B file: D:\V-Tracks-48-nero_extreme.wav

00:59:25 1/1 p=50.0%
00:59:45 2/2 p=25.0%
01:00:31 2/3 p=50.0%
01:00:50 2/4 p=68.8%
01:00:52 reset

01:01:10 1/1 p=50.0%
01:01:51 2/2 p=25.0%
01:05:28 2/3 p=50.0%
01:05:30 reset

01:05:39 1/1 p=50.0%
01:06:35 1/2 p=75.0%
01:06:37 reset

01:06:45 1/1 p=50.0%
01:06:53 2/2 p=25.0%
01:07:02 3/3 p=12.5%
01:07:47 3/4 p=31.2%
01:09:23 4/5 p=18.8%
01:10:31 4/6 p=34.4%
01:10:34 reset

01:10:54 1/1 p=50.0%
01:11:03 2/2 p=25.0%
01:11:12 2/3 p=50.0%
01:11:15 reset

01:11:24 1/1 p=50.0%
01:11:37 1/2 p=75.0%
01:11:38 reset

01:11:55 1/1 p=50.0%
01:13:18 1/2 p=75.0%
01:13:20 reset

01:13:30 1/1 p=50.0%
01:14:09 1/2 p=75.0%
01:14:13 reset

01:17:41 1/1 p=50.0%
01:20:06 1/2 p=75.0%
----------------------------------------

I can always detect the "swoosh" sound on my first ABX attempt, but I get tired after.

SK1 had much more luck. He uses a better sound card, an Audigy, as opposed to my SB128
with its weak volume. His headphones are more suited for such a test (Sennheiser HD580,
while I still have the open HD495S). Lucky him.

Here are the logs from his various attempts:

LAME 3.90.3 --alt-preset insane (easy):

WinABX v0.41 test report
02/05/2004 21:25:17

A file: E:\test\V-Tracks-48.wav
B file: E:\test\V-Tracks-48-lame_api.wav

Start position 00:00.0, end position 00:08.2
21:25:39 1/1 p=50.0%
21:25:57 2/2 p=25.0%
21:26:20 3/3 p=12.5%
21:26:40 4/4 p=6.2%
21:27:21 5/5 p=3.1%
21:27:29 6/6 p=1.6%
21:27:51 7/7 p=0.8%
21:28:09 8/8 p=0.4%
21:28:45 9/9 p=0.2%
21:28:54 10/10 p< 0.1%
21:29:07 11/11 p< 0.1%
21:29:22 12/12 p< 0.1%
21:29:24 test finished
------------------------

Nero 6.3.0.3, Extreme, VBR. Both Nero samples were kindly prepared for us by menno. (very easy):

WinABX v0.41 test report
02/06/2004 00:51:07

A file: E:\test\V-Tracks-48.wav
B file: E:\test\V-Tracks-48-nero_extreme.wav

Start position 00:00.0, end position 00:08.2
00:52:04 1/1 p=50.0%
00:52:12 2/2 p=25.0%
00:52:24 3/3 p=12.5%
00:52:37 4/4 p=6.2%
00:52:46 5/5 p=3.1%
00:52:55 6/6 p=1.6%
00:53:11 7/7 p=0.8%
00:53:24 8/8 p=0.4%
00:53:32 9/9 p=0.2%
00:53:39 10/10 p< 0.1%
00:53:44 11/11 p< 0.1%
00:53:50 12/12 p< 0.1%
00:54:03 test finished
------------------------

Nero 6.3.0.3, Audiophile, VBR (very hard):

WinABX v0.41 test report
02/06/2004 01:26:45

A file: E:\test\V-Tracks-48.wav
B file: E:\test\V-Tracks-48-nero_audiophile.wav

Start position 00:00.0, end position 00:08.2
01:27:23 1/1 p=50.0%
01:27:30 2/2 p=25.0%
01:27:59 3/3 p=12.5%
01:28:09 4/4 p=6.2%
01:28:20 5/5 p=3.1%
01:29:49 6/6 p=1.6%
01:31:55 7/7 p=0.8%
01:32:08 test finished
-----------------------

Oggenc GT3b2, Xiph.Org/Sjeng.Org libVorbis I 20030909 (GTune 3, beta 2), -q 8 (hard):

WinABX v0.41 test report
02/06/2004 01:45:08

A file: E:\test\V-Tracks-48.wav
B file: E:\test\V-Tracks-48-vorbis_q8.wav

Start position 00:00.0, end position 00:08.2
01:45:30 1/1 p=50.0%
01:45:44 2/2 p=25.0%
01:46:25 3/3 p=12.5%
01:46:36 4/4 p=6.2%
01:47:04 5/5 p=3.1%
01:47:55 6/6 p=1.6%
01:48:29 7/7 p=0.8%
01:48:36 test finished
-----------------------

DualStream --mode extra --quality 0, file bitrate = 276.8 kbps (failure to ABX):

WinABX v0.41 test report
02/06/2004 02:18:39

A file: E:\test\V-Tracks-48.wav
B file: E:\test\v-tracks-48-ofs_extra_quality0.wav

Start position 00:00.0, end position 00:08.2
02:18:57 1/1 p=50.0%
02:19:20 2/2 p=25.0%
02:19:39 2/3 p=50.0%
02:19:48 2/4 p=68.8%
02:20:21 3/5 p=50.0%
02:20:34 4/6 p=34.4%
02:20:42 4/7 p=50.0%
02:20:47 test finished
-----------------------

MPC - mppenc 1.15r --quality 7 (failure to ABX):

WinABX v0.41 test report
02/06/2004 02:23:34

A file: E:\test\V-Tracks-48.wav
B file: E:\test\V-Tracks-48-mpc_insane.wav

Start position 00:00.0, end position 00:08.2
02:24:18 0/1 p=100.0%
02:24:54 1/2 p=75.0%
02:25:20 2/3 p=50.0%
02:25:58 3/4 p=31.2%
02:26:21 3/5 p=50.0%
02:27:03 4/6 p=34.4%
02:27:24 4/7 p=50.0%
02:28:03 5/8 p=36.3%
02:28:22 6/9 p=25.4%
02:28:52 6/10 p=37.7%
02:29:36 6/11 p=50.0%
02:29:45 test finished
------------------------

We would like to hear about your experiences too. With AAC, mostly, since it's
relevant to this forum. I am adding a link to the original sample, APE, 48 KHz
(no need to SSRC) through a .torrent file.

===========================================

V-Tracks - Subway 26 - Monkey's Audio, extra high, 48 KHz

This post has been edited by Seed: Feb 6 2004, 03:28

[Supachikn]

WinABX v0.41 test report
02/08/2004 10:06:08

A file: E:\sync-48.wav
B file: E:\sync-48-nero_extreme.wav

Start position 00:00.0, end position 00:13.6
10:06:33 1/1 p=50.0%
10:06:42 1/2 p=75.0%
10:07:03 2/3 p=50.0%
10:07:13 3/4 p=31.2%
10:07:29 3/5 p=50.0%
10:07:46 4/6 p=34.4%
10:07:54 5/7 p=22.7%
10:08:09 6/8 p=14.5%
10:08:28 7/9 p=9.0%
10:08:43 8/10 p=5.5%