I found 2 damaged mp3. There are clicks after about 50% in one and white noise in the other after 40%. The first was grabbed from my CD. I grabbed it once again (got same CRC) and encoded - OK. But the second is the rare one. I can't get that CD again. I am sure that CD was OK. But I didn't listen MP3 after encoding. After encoding these files didn't moved over the hardrive. Just burned after encoding. Please help!

Is there possibility to repair them?

AND WHY DID THIS HAPPEND??????????? ?

Unfortunately you can't "repair" the mp3 without the original CD after it's been encoded... the best you can do is try to edit out some clicks or bad portions in a wav editor like Cooledit or Soundforge before you burn it to CD again (otherwise you'll then have to transcode it back to mp3, as wav editors obviously don't edit mp3s... transcoding would make the file sound much worse). However, that's hardly "fixing" it. You'll need to get the original CD back and re-rip it properly.

Clicks are often caused by scratches or dirt on the CD, or sometimes by a bad ripper. You should use EAC (www.exactaudiocopy.de) to rip, and configure it according to the guides and posts present here (do a search).

I don't know how white noise not present in the original would occur... what encoder are you using? See the sections on "Recommended LAME compiles" and "Recommended LAME settings" if you're in doubt as to which encoder version and/or settings to use.

your hard drive could be starting to fail, before thinking about these mp3s you'd better backup all your important data and run some extensive HD tests!

Grabbing - EAC
Encoding - LAME.exe APS, APS -Z
HD seems to be OK, but I'll try more extensive tests.
As I understand this couldn't be lame.exe fault?

well, if the file was once good and then misteriously failed, It's VERY likely it's your HD, considering you haven't touched the file.

I would have to agree with /\/ephaestous, I can't imagine what else would cause the problems you're describing...

It's can be bad soundcard. Try to play this file on other computer and see.

You should also try some other player like foobar2000 if you are using Winamp.

What filesystem did you use for ripping the CD/storing the resulting mp3 afterwards? If one of them was FAT based there's a chance than some random error (OCed system, misbehaved programs in 98 or even NT based Win) corrupted the FAT tables and/or file data.

Once I had a very nasty crash in Win2k while the the WinAmp was playing. After reboot Win did make some corrections to the drive which holds the swap file but didn't even check the drive with the mp3s. The strange thing was that the file which was playing in the winamp at the moment of crash got some frames mixed up with another mp3!!!
I guess this is just one of those rare occurances but it makes you wonder...

Thanks for all your replies. Here are answers on your questions.
I made hard test on my HD with Norton Utilities. No errors found, neither bad blocks.
I would be happy if bad blocks appeared. The problem would be found. But now
If the problem was once I would stay cool. But there are two MP3s.
I played the broken ones on iRiver SlimX and using WinAmp.
My system is WIN98SE and HD is FAT32.

However I MP3Gained that broken MP3s. One of them come with >100db, the second one can't be analyzed by MP3Gain.
After scanning it I get result of '???' And it can't be gained.
So I am using MP3 Gain to find out other broken files now.

Is there any any other way to find out other broken ones?
What is CRC correction? Anybody used it?
Will LAME with CRC make larger files or the size will remain the same and quality will be lower (with APS)?
Can CRC info be deleted before burning on CD?

Definitely, there are broken frames in that mp3. The way it has been broken is what remains unclear.

As of a tool to identify broken mp3 (and tips to fix them), you can use Mp3Utility http://www.geocities.com/mp3utility

The question is.... in the 2nd file, the white noise persists,or just happens there? If that part is damaged, you could remove that frame, and "problem solved" (not really, but a small skip wouldn't be as noticeable as that). If the white noise persists, then there's either bad data there (not at all mp3 data), or there's a resincronization with a different layer, frequency or something ( mp3utility will inform you of this if I remember well )


About CRC checksum, a) size remains the same, which implies less space for real data, which implies less quality. Next, it is just a detective mechanism, not a proactive one. This means that only allows to detect problems, not to fix them. It is of little use for storage purposes.

Thanks for that utility, I downloaded it and will try.
About white noise. There are no music after about 40% of song. Only that white noise. The lenght of track is correct. Even when it is in it's album folder I can CDDB album.
The presence of white noise seems very strange for me. I did some tests. I edited some MP3s. Just cut out or inserted some data from (in) them. Always get file with skips. And never white noise.
Any ideas how must look data in mp3 to make that white noise?

Thanks again for that utility, I will scan all my collection. But. There are no errors in the broken one (with white noise) regarding utility. All my edited (cuted with notepad) comes with errors. One more test. I tried not to cut MP3 but to change some digits and letters with others. And that doesn' gives error. So that utility doesn't give 100% warranty .


And after all tests (WAV CRC is the same, HD scan, cutting MP3s...) it seems to me that LAME somehow encoded wrong. What do you think about that. Is it possible.

You could try to use MD5 Summer and calculate the checksum for the files.
When you hear unexpected clicks on the files then check their checksums to see if something it's changed.

Don't use notepad to edit binary files. you'll always end loosing data. Use any hex editor avaiable, here there are the links to two freeware applications that can edit binary files

http://www.hhdsoftware.com/hexeditor.html (HDD hex editor. An explicit hex editor)
http://www.pnotepad.org/ (Programmers' notepad. An all in one, text-colouring editor)



That utility validates the frames in which the mp3 data is stored. it does not validate the data itself, since it can be anything.
You can think an mp3 file is like a group of packets which have a label, and inside, there's the data. The problem appears when data is missing and a label can be interpreted as part of data, or viceversa.



There's only one question at this point: Is your PC overclocked?

It could happen that one bit was skipped at some point when *reading* the file, which would have produced the wave being encoded to be completely wrong.
(Aside of this, I think LAME had a few issues with overclocked PC's)

Edit: Made quotes work

1. When used notepad I wanted to create errors in the file
3. The strange thing is demolishing MP3s didn't make any white noise. Only clicks and skips.
2. CPU isn't overclocked, however I think you are right. It's not HD. It seem to be CPU. I am going paranoid with this. Thinking about encoding rare CD's twice and compare them encoded files. And of course will use MD5(but in this case MP3 was broken just after encoding)
Thank you very much.

P.S. Where can I download MD5?

Here:
http://www.md5summer.org/download.html