Lame 3.97b1 vs Original CD
Reply #22 – 2005-09-30 18:47:52
I'm oversensitive to high frequency transients but not lucky about it. Thats why i usually avoid taking part in codec-tests, because i dont want to *train* myself in noticing even "better" what annoys me already anyways. But after listening to this sample, i thought that this might be a quick'n easy job for me where i could help out. I encoded the sample with LAME 3.97b1 to preset fast medium, used foobars ABX-comparator, and did hide the results, so that i wouldn't need to decide about the number of trials beforehand (so that i could stop when suffering from fatigue which i knew would happen soon with this kind of audiomaterial). I stopped after 7 trials, because i felt that the difference was beginning to fade and that my ears began to imagine all kinds of things.... the results after i did unhide them were:foo_abx 1.2 report foobar2000 v0.9 beta 8 2005/09/30 19:29:54 File A: D:\clip.wav File B: D:\2 clip.mp3 19:29:55 : Test started. 19:31:52 : 01/01 50.0% 19:32:34 : 02/02 25.0% 19:34:11 : 03/03 12.5% 19:35:05 : 04/04 6.3% 19:36:35 : 05/05 3.1% 19:37:55 : 05/06 10.9% 19:41:15 : 06/07 6.3% 19:49:27 : Test finished. ---------- Total: 6/7 (6.3%) edit: i focussed on the short and sharp attacks. They have pre-echo issues in the LAME-version, which doesn't suprise me that much, because those drums are almost microattacks. edit2: kbit-average for the APFM-mp3 is 172