Hello
I've decided to test the new version at cbr 128. I'm particularly interested in this setting because a website where I upload my songs requires the mp3's to be cbr, and since the offer limited space I want the files to be as small as possible. While testing lame_june against lame 3.97 alpha 10 (compiled from cvs, LAME version 3.97 (alpha 10, Apr 8 2005 12:38:33)) and lame 3.96.1 from Rarewares with this sample (from Infected Mushroom - Never Ever Land) I noticed an annoying glitch. It appears at 0:12.15 in files encoded with alpha 10 and lame_june, but not 3.96.1. Here's the log:
ABC/HR for Java, Version 0.5a, 06 czerwiec 2005
Testname: im_01
Tester:
1R = D:\render\test\im_01_a10.wav
2L = D:\render\test\im_01_june.wav
3L = D:\render\test\im_01_3961.wav
---------------------------------------
General Comments:
---------------------------------------
1R File: D:\render\test\im_01_a10.wav
1R Rating: 2.5
1R Comment:
---------------------------------------
2L File: D:\render\test\im_01_june.wav
2L Rating: 2.5
2L Comment:
---------------------------------------
3L File: D:\render\test\im_01_3961.wav
3L Rating: 3.5
3L Comment:
---------------------------------------
ABX Results:
D:\render\test\im_01_june.wav vs D:\render\test\im_01_3961.wav
8 out of 8, pval = 0.0030
D:\render\test\im_01_a10.wav vs D:\render\test\im_01_june.wav
5 out of 8, pval = 0.363
---- Detailed ABX results ----
D:\render\test\im_01_june.wav vs D:\render\test\im_01_3961.wav
Playback Range: 11.688 to 19.832
7:02:19 PM p 1/1 pval = 0.5
7:02:22 PM p 2/2 pval = 0.25
7:02:25 PM p 3/3 pval = 0.125
7:02:27 PM p 4/4 pval = 0.062
7:02:30 PM p 5/5 pval = 0.031
7:02:33 PM p 6/6 pval = 0.015
7:02:35 PM p 7/7 pval = 0.0070
7:02:38 PM p 8/8 pval = 0.0030
D:\render\test\im_01_a10.wav vs D:\render\test\im_01_june.wav
Playback Range: 05.514 to 19.832
6:54:32 PM f 0/1 pval = 1.0
6:54:35 PM f 0/2 pval = 1.0
Playback Range: 11.688 to 19.832
6:55:31 PM p 1/3 pval = 0.875
6:55:34 PM f 1/4 pval = 0.937
6:55:38 PM p 2/5 pval = 0.812
6:55:44 PM p 3/6 pval = 0.656
6:55:50 PM p 4/7 pval = 0.5
6:55:54 PM p 5/8 pval = 0.363
I tried abr 128 (and it solves the glitch problem) too, but I failed at ABX'ing different encodes from one another.
I did another test with another sample (from Infected Mushroom - B.P. Empire). The hihats seem very watery and there's some noticeable ringing (is this the right word?). What's interesting, lame_june sounds better for me than alpha 10, and I failed to ABX it from 3.96.1.
The log:
ABC/HR for Java, Version 0.5a, 06 czerwiec 2005
Testname: im_02
Tester:
1R = D:\render\test\im_02_3961.wav
2R = D:\render\test\im_02_june.wav
3L = D:\render\test\im_02_a10.wav
---------------------------------------
General Comments:
---------------------------------------
1R File: D:\render\test\im_02_3961.wav
1R Rating: 3.0
1R Comment:
---------------------------------------
2R File: D:\render\test\im_02_june.wav
2R Rating: 3.0
2R Comment:
---------------------------------------
3L File: D:\render\test\im_02_a10.wav
3L Rating: 2.0
3L Comment:
---------------------------------------
ABX Results:
D:\render\test\im_02_june.wav vs D:\render\test\im_02_a10.wav
8 out of 8, pval = 0.0030
D:\render\test\im_02_3961.wav vs D:\render\test\im_02_a10.wav
8 out of 8, pval = 0.0030
---- Detailed ABX results ----
D:\render\test\im_02_june.wav vs D:\render\test\im_02_a10.wav
Playback Range: 15.111 to 33.653
9:16:06 PM p 1/1 pval = 0.5
9:16:09 PM p 2/2 pval = 0.25
9:16:14 PM p 3/3 pval = 0.125
9:16:16 PM p 4/4 pval = 0.062
9:16:19 PM p 5/5 pval = 0.031
9:16:22 PM p 6/6 pval = 0.015
9:16:25 PM p 7/7 pval = 0.0070
9:16:31 PM p 8/8 pval = 0.0030
D:\render\test\im_02_3961.wav vs D:\render\test\im_02_a10.wav
Playback Range: 15.111 to 33.653
9:14:52 PM p 1/1 pval = 0.5
9:14:54 PM p 2/2 pval = 0.25
9:14:56 PM p 3/3 pval = 0.125
9:14:59 PM p 4/4 pval = 0.062
9:15:04 PM p 5/5 pval = 0.031
9:15:07 PM p 6/6 pval = 0.015
9:15:10 PM p 7/7 pval = 0.0070
9:15:13 PM p 8/8 pval = 0.0030
I'll try to do some more tests as soon as I gain access to my other CDs.
edit: following ChiGung's suggestion changed code, to codeboxes, so the post isn't that long