LAME 3.97 320kbps CBR
Default vs -q 2
foo_abx 1.3.4 report
foobar2000 v1.0.1
2010/03/20 20:00:23
File A: C:\Temp\Show Me Your Spine 320 CBR.mp3
File B: C:\Temp\Show Me Your Spine 320 CBR -q 2.mp3
20:00:23 : Test started.
20:00:52 : 01/01 50.0%
20:00:59 : 02/02 25.0%
20:01:07 : 03/03 12.5%
20:01:13 : 04/04 6.3%
20:01:22 : 05/05 3.1%
20:01:26 : 06/06 1.6%
20:01:35 : 07/07 0.8%
20:01:45 : 08/08 0.4%
20:01:51 : 09/09 0.2%
20:02:09 : 10/10 0.1%
20:02:14 : 11/11 0.0%
20:02:19 : 12/12 0.0%
20:02:20 : Test finished.
----------
Total: 12/12 (0.0%)
The obivous pre-echo artifact at the very start improves greatly when using the -q 2 switch. However there is a encoding speed decrease (16x with q 2 and 25x with the default setting, on a Core 2 Duo E6600 CPU).
Using -q 2 on this sample is still far from transparent:
foo_abx 1.3.4 report
foobar2000 v1.0.1
2010/03/20 20:02:48
File A: J:\Music\Lossless\Ministry - Side Trax\13. Show Me Your Spine (PTP).flac
File B: C:\Temp\Show Me Your Spine 320 CBR -q 2.mp3
20:02:48 : Test started.
20:03:01 : 01/01 50.0%
20:03:07 : 02/02 25.0%
20:03:15 : 03/03 12.5%
20:03:22 : 04/04 6.3%
20:03:38 : 05/05 3.1%
20:03:51 : 06/06 1.6%
20:04:03 : 07/07 0.8%
20:04:14 : 08/08 0.4%
20:04:29 : 09/09 0.2%
20:04:35 : 10/10 0.1%
20:04:44 : 11/11 0.0%
20:04:52 : 12/12 0.0%
20:04:53 : Test finished.
----------
Total: 12/12 (0.0%)
I have also tried out -q 0 with a more pre-echo prone sample:
foo_abx 1.3.4 report
foobar2000 v1.0.1
2010/03/20 19:42:58
File A: C:\Temp\musique-non-stop 320 CBR.mp3
File B: C:\Temp\musique-non-stop 320 CBR -q 0.mp3
19:42:58 : Test started.
19:43:33 : 01/01 50.0%
19:43:55 : 02/02 25.0%
19:44:14 : 03/03 12.5%
19:44:41 : 04/04 6.3%
19:45:16 : 04/05 18.8%
19:45:26 : 05/06 10.9%
19:45:35 : 06/07 6.3%
19:45:57 : 07/08 3.5%
19:46:13 : 08/09 2.0%
19:46:35 : 09/10 1.1%
19:46:42 : 10/11 0.6%
19:46:50 : 11/12 0.3%
19:46:57 : 12/13 0.2%
19:47:06 : 13/14 0.1%
19:47:14 : 14/15 0.0%
19:47:26 : 15/16 0.0%
19:47:35 : 16/17 0.0%
19:47:47 : 17/18 0.0%
19:47:54 : Test finished.
----------
Total: 17/18 (0.0%)
This sample is just riddled with pre-echo that even -q 0 barely makes any improvement. The only noticable improvement was the artifact at 0:05 was clear on the -q 0.
Also the speed for -q 0 is pretty bad, i got 10x - 11x and IMO not really worth the small improvement.