data deterioration

[tazcarper]

My friend and I were having a debate about data deterioration. The claim, which I disagreed with, is that if a .mp3 file is transferred from one computer to another about a thousand times, the quality of the mp3 will deteriorate. So if a mp3 file started on computer 1 and then went to 2, then 2 transferred it to 3 and 3 transferred it to 4 (ect ect...), by the time you got to computer 999 transferring it to 1000, the mp3 quality will be worse.

To my understanding, the quality of the mp3 file, or any file, will always be consistent when transferred unless there is a hardware failure, transfer failure, the file is compressed or there is 3rd party interference.

Can some of you fine folks please help explain to me if I am correct or if I am wrong. The only thing I could think of to counter this claim would be "bit rot", but bit rot (i thought) was due to outdated software or hardware failure.

Any help would rock!
Thank you in advance.

 

[phide]

If you transfer a file from one machine to another, and the transfer is not entirely successful, Windows notifies you. This is the same in other operating systems, and this is true of any file -- compressed or uncompressed (they're all represented identically).

To answer your question, no, the file does not deteriorate unless there is an error. When there's an error, you'll be notified of it. It isn't simply something you'll come across some time later.

With MP3s specifically, if the transferred file is identical, the quality is identical. Audio files certainly are no different than any other file with respect to degradation: they will never degrade.

 

[evilsofa]

A checksum is used to check whether a file has been corrupted during a copy. Checksums were developed at the very beginning of the digital age due to dire necessity, to prevent just what you are concerned about.

http://en.wikipedia.org/wiki/Checksum

In the chain you speak of, it's quite possible that somewhere along the line, a computer with bad hardware (especially bad RAM) might send a corrupted version of the .mp3 to the next computer, but the corrupted .mp3 would not work at all on that next computer; it would just be an unusable file. The quality of the file would not degrade - it would either work perfectly or not work at all.

The only way in which it would degrade as it goes down the chain is if each user in the chain converts the file from one format to another - say, converting .mp3 to .wmv and back again. It's worth noting that .mp3 is already degraded to begin with - it's not a lossless format. But the first .mp3 in the chain will either be exactly identical to the 1,000th copy or it will not work at all.

 

[unhappy_mage]

In the chain you speak of, it's quite possible that somewhere along the line, a computer with bad hardware (especially bad RAM) might send a corrupted version of the .mp3 to the next computer, but the corrupted .mp3 would not work at all on that next computer; it would just be an unusable file. The quality of the file would not degrade - it would either work perfectly or not work at all.

Not true. You'd get a glitch in the audio, that's all.

 

[Eva_Unit_0]

Copying data on a computer isn't like using a photocopier where there is a clear difference between "originals" and "copies." There should be no difference at all between file copies. Barring some sort of hardware problem, every copy of a file should be bit-identical to the original. And as others have stated, pretty much all modern operating systems do checksums on copies to ensure that they are indeed identical. You'll get CRC errors if they aren't.

If a bad piece of hardware causes memory corruption or something you could see negative effects on the file, but that's due to hardware failure, not due to some inherent limitation of how file copying works.

I'll write up a bash script to copy a file a zillion times and show that they're the same.

 

[Eva_Unit_0]

Ok here's what I did. I wrote a script to copy an mp3 (since that is the particular file type in question, though it doesn't really matter) 500 times between two disks in my system. The song I chose was Immigrant Song by Led Zeppelin. I actually had to think for a while about which song I wanted to use. Here's the script:

#!/bin/bash
COUNTER=0
while [ $COUNTER -lt 500 ]; do
	echo copy number $COUNTER
	mv /root/08\ Immigrant\ Song.mp3 /home/keith/08\ Immigrant\ Song.mp3
	sync
	echo "first copy done"
	mv /home/keith/08\ Immigrant\ Song.mp3 /root/08\ Immigrant\ Song.mp3
	sync
	echo "second copy done"
	let COUNTER=COUNTER+1
done

/ and /home are on separate drives. Yes, I realize that I'm flagrantly abusing 'sync' here. I had to, otherwise it would just do all the copying and deleting in the disk buffer and then sync at the end. So yes, we can be confident that it actually copied the file 500 times and didn't just "fake" it.

Here is the md5 of the file at the start:

hamlet ~ # md5sum 08\ Immigrant\ Song.mp3 
ca13cdb55e1fef0c9aeeaad42fe8e32c  08 Immigrant Song.mp3

Here is the md5 after 500 copies:

hamlet ~ # md5sum 08\ Immigrant\ Song.mp3 
ca13cdb55e1fef0c9aeeaad42fe8e32c  08 Immigrant Song.mp3

It's the same! This means that the files must sound identical. There is no possible way it could sound "deteriorated" if it is still bit-identical to the original.

 

[phide]

Not true. You'd get a glitch in the audio, that's all.

If the header were corrupted, it's possible the file wouldn't be recognizable in a media player.

 

[Demon10000]

It's the same! This means that the files must sound identical. There is no possible way it could sound "deteriorated" if it is still bit-identical to the original.

Of course!

Your computer doesn't distinguish between music, data, documents... it seems them a data on a disk. If you were to copy an executable file over and over, you would expect the end result to be the same, correct? And MP3 file is no different.

You thinking in about analog transfers -- like copying from tape to tape to tape to tape... not CD to CD.

 

[tazcarper]

Thanks for the fast responses. Checksum was what I was trying to think of when we were debating the issue. I knew there was something that verified the data, but couldn't remember what it was called.

Thanks again!

 

[LookTJ]

Also the only way an mp3 quality degrades, is when it is transcoded into another lossy format such as ogg.