Help - Search - Members - Calendar
Full Version: Lame 3.96.1 Wrong Testcase.mp3 ?
Hydrogenaudio Forums > Lossy Audio Compression > MP3 > MP3 - Tech
mykhal
Feb 8 2005, 00:37
I thought "make test" in lame sources is designed to check whether custom compiled code produces exactly the same binary result as the executable prepared by developers. But it seems, that it serves for comparison of the output with previus lame release's output, and does not tell anything about the difference between my executable, and the same version `reference' executable. Or have developers forgotten to update the sample ?

My testcase.mp3 (md5sum: c731431aa91edde7422c03215ed23b77) is 10425b long, and testcase.new.mp3 has 10448b. cmp produces 8905 differences (independent on compiler CFLAGS), it means the mp3s differ in approximately 85% bytes! (I know I shouldn't count bytes, but rather differing bits, which would lead into smaller number)

It seemed too much for me, so I looked into files:

CODE

<------testcase.mp3------> <----testcase.new.mp3---->
0000000: fffb904400000000 ...D.... ...D.... fffb904400000000
0000008: 0000000000000000 ........ ........ 0000000000000000
0000010: 0000000000000000 ........ ........ 0000000000000000
0000018: 0000000000000000 ........ ........ 0000000000000000
0000020: 00000000496e666f ....Info ....Info 00000000496e666f
0000028: 0000000f00000018 ........ ........ 0000000f00000018
0000030: 000028b9000a0a0a ..(..... ..(..... 000028d0000a0a0a
0000038: 0a15151515202020 ..... ..... 0a15151515202020
0000040: 202a2a2a2a353535 ****555 ****555 202a2a2a2a353535
0000048: 3540404040404a4a 5@@@@@JJ 5@@@@@JJ 3540404040404a4a
0000050: 4a4a555555556060 JJUUUU`` JJUUUU`` 4a4a555555556060
0000058: 60606a6a6a6a7575 ``jjjjuu ``jjjjuu 60606a6a6a6a7575
0000060: 757580808080808a uu...... uu...... 757580808080808a
0000068: 8a8a8a95959595a0 ........ ........ 8a8a8a95959595a0
0000070: a0a0a0aaaaaaaab5 ........ ........ a0a0a0aaaaaaaab5
0000078: b5b5b5c0c0c0c0c0 ........ ........ b5b5b5c0c0c0c0c0
0000080: cacacacad5d5d5d5 ........ ........ cacacacad5d5d5d5
0000088: e0e0e0e0eaeaeaea ........ ........ e0e0e0e0eaeaeaea
0000090: f5f5f5f5ffffffff ........ ........ f5f5f5f5ffffffff
0000098: 000000394c414d45 ...9LAME ...9LAME 000000394c414d45
00000a0: 332e39352001af00 3.95 ... 3.96r... 332e39367201af00
00000a8: 0000002e4c000014 ....L... ........ 0000002e10000014
00000b0: 802408186d000000 .$..m... .$..n... 802408186e000080
00000b8: 000028b91a26ff67 ..(..&.g ..(.i.;B 000028d069963b42
00000c0: 0000000000000000 ........ ........ 0000000000000000
...
many differences till the end


Do you see it too ?
Gabriel
Feb 9 2005, 09:32
You are right, we forget to update it for 3.96.1.
Yogiboar
Feb 9 2005, 09:53
QUOTE (Gabriel @ Feb 9 2005, 12:32 AM)
You are right, we forget to update it for 3.96.1.
*



So what is the impact [if any] for less knowledgable users? unsure.gif
westgroveg
Feb 9 2005, 10:28
QUOTE (Yogiboar @ Feb 9 2005, 08:53 PM)
QUOTE (Gabriel @ Feb 9 2005, 12:32 AM)
You are right, we forget to update it for 3.96.1.
*



So what is the impact [if any] for less knowledgable users? unsure.gif
*


This won't affect quality. Just use the rarewares LAME executables.
mykhal
Feb 9 2005, 21:55
QUOTE (Gabriel @ Feb 9 2005, 12:32 AM)
You are right, we forget to update it for 3.96.1.
*

Thanks for confirmation. I was unsure what's wrong.
So is it possible to get it somewhere right now ?

QUOTE (Yogiboar @ Feb 9 2005, 12:53 AM)
So what is the impact [if any] for less knowledgable users?  unsure.gif
*

For less knowledgable users it does not affect anything. Wrong is just the testing sample, not the program itself.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2009 Invision Power Services, Inc.