I just listened to music with my headphone quite load using foobar and everything was fine. I played a other song I hasn´t played yet with foobar. I noticed it sounds a bit strange comapared to winamp, I thought that´s just imagination, but then there was a very load noise in it for about for about 0,1 sec. Because I listened the music load as I used to do, that noise, which was much loader, produced such a pain in my ear I nearly kicked in my pc in that rage.
After about 10 minutes when the pain was gone and just a this noise in the ears when you hear something to load was left I played the file in again (without headphones and quietly). Excact at the same place the same strange noise again. I played the file in winamp (2.81), just fine. I disabled the mad-plug-in and played it, also fine. I enabled mad again and I saw there up to this point 3 sync errors and 8 other errors in the mp3, I never noticed that because it always played normal. In wmp it also plays normal.
But anyway no decoder should produce that noise. Sometimes I listen to music even louder, I could have been deaf now.
I uploaded a part of the mp3, the noise comes at 0:23: Respect.mp3 (660KB)
I suggest you use an other decoder. Until that isn´t fixed I´ll switch back to winamp + mad.

This semms to me not as a decoder error, decoder of Foobar works just fine, it has not that good error-correction like winampdecoder, in winamp (2.81, no 3-rth party plugins, all standart) the sound definetly crackeld at 0.23, it is a file-error in the mp3 file. if you don't hear any error, i think, then your decoder-version is an other than mine and is correction this error on-the-fly. I can her the error in Winamp and Mediaplayer!

ok, dloading the file, i'll check it next time i mess with mpglib.
its most likely another broken file, mpglib bad data safety is still awful.
on a side note, i have some damaged files that sound less annoying in fb2k than in winamp :P

It´s not a decoder error not play this file (it could stop, crash etc.) but to output that is defintly a bug. The mp3-decoder in winamp is Nullsoft MPEG Audio Decoder 2.81 and there is a very short silence, but not that crack. In wmp (Fhg-mp3-decoder 1.5 build 50) the same. In mad it plays fine, because it has a very good error correction.

Edit:
It is a broken file (mad says that in statistics), but I never noticed that with the good error-handling of the mad-decoder. mpglib should better crash or stop playback when reading corrupt data.

On tests that I've seen mad v 12 does not decode properly I am not sure if these problems were fixed in v 14. Mpglib on the other hand passed all portions of the test.

(With Mp3utility)

***
Processing: C:\WINDOWS\Desktop\Respect.mp3

Error: Sync error reading frame header 105 expected at byte 43,468. Approx. time: 0:03 (6% through audio).
Resync successful - Frame header 105 found at byte 43,220.
Frame 104 (bytes 43,050 - 43,219) short by 248 bytes (expected 418 bytes, found 170 bytes).

Error: Sync error reading frame header 375 expected at byte 156,069. Approx. time: 0:10 (23% through audio).
Resync successful - Frame header 375 found at byte 155,899.
Frame 374 (bytes 155,651 - 155,898) short by 170 bytes (expected 418 bytes, found 248 bytes).

Error: Sync error reading frame header 899 expected at byte 374,910. Approx. time: 0:23 (55% through audio).
Resync failed (no matching frame header found within 2,000 bytes).

Summary: 898 total frames processed (861 padded, 37 unpadded). Bitrate is constant.
***

Here is what madplay (self-compiled, don´t know where you can find binary) reported:

F:\>madplay -v respect.mp3
MPEG Audio Decoder 0.14.2 (beta) - Copyright © 2000-2001 Robert Leslie
00:00:02 Layer III, 128 kbps, 44100 Hz, joint stereo (LR), CRC
error: frame 104: lost synchronization
00:00:09 Layer III, 128 kbps, 44100 Hz, joint stereo (LR), CRC
error: frame 373: lost synchronization
00:00:23 Layer III, 128 kbps, 44100 Hz, joint stereo (LR), CRC
error: frame 897: bad scalefactor selection info
00:00:42 Layer III, 128 kbps, 44100 Hz, joint stereo (MS), no CRC
1610 frames decoded (0:00:42.0), -6.7 dB peak amplitude, 0 clipped samples

Playback works fine. The file is defintly broken, but that shouldn´t cause this teribble strange cracking, just a small pause for that frame (~0,026 sec.)