Welcome to the MacNN Forums.

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

You are here: MacNN Forums > Community > MacNN Lounge > MP3 Degradation Over Time

MP3 Degradation Over Time
Thread Tools
Mac Elite
Join Date: Jul 2001
Status: Offline
Reply With Quote
Feb 15, 2003, 01:40 PM
 
Over the past four years, I have many times transferred the constantly-growing number of MP3 files I have back and forth between various hard disks. These disks are in an Rev D iMac 500, a Pismo 400, a ClubMac USB hard drive ( n: ), a Maxtor FW HD ( ) , and my current baby, a LaCie FW HD.

All my MP3s are iTunes rips at 192 kpbs from my CDs, spanning several versions of iTunes, Mac OSes, and computers. I believe that they are all functionally equivalent - that is, that the ripping process is the same with modern iTunes and that it plays old ones just fine. That's the point of having a standard, yes?

In any case, I've noticed recently that, starting with the older ones, I'm getting short periods of silence at random times in songs. These are between .5 seconds and 5 seconds. These occur randomly, as far as I can tell, but only in the MP3s which I have had in the collection for at least 2 years. I have also noticed that the track info has random characters in it once in a while, or track ID info switched (always between adjacent tracks on an album). And yes, it'[s consistent; if I try to play the same song again (maybe iTunes made an error the last time it played), the silence is in the same place, forever and ever, amen.

Does anyone have a ballpark figure for the accuracy of reading and writing on modern hard disks? If there's one error every million bytes transferred, and I have transferred several GBs dozens and dozens of times, there's likely to be these odd things cropping up, yes?

Is there a verifying part of the writing process? I mean, read data from A, write to B, read B and compare to A? I mean, would this be something basic, handled by hardware, or would the OS have a look?

It won't matter in a couple years, or perhaps even sooner - I think that I will delete my MP3s and start over (got everything on CDs in a box in my closet), but doing some serious trimming of what I don't listen to. This time, I'll import to AIFF ( that's just going to be a copy of the CD audio files ) and make sure that there's something always checking and comparing files copied. The limiting factor is my 20 GB iPod - my storage space is fine otherwise.

It's disconcerting, though - what if I actually had something important stored digitally, not on paper?
Actual conversation between UCLA and Stanford during a login on early Internet - U: I'm going to type an L! Did you get an L? S: I got one-one-four. L! U:Did you get the O? S: One-one-seven. U: <types G> S: The computer just crashed.
     
Professional Poster
Join Date: Aug 2002
Location: Floreeda
Status: Offline
Reply With Quote
Feb 15, 2003, 06:59 PM
 
i want AAC
     
Posting Junkie
Join Date: Jun 2001
Location: Baltimore, MD
Status: Offline
Reply With Quote
Feb 15, 2003, 07:14 PM
 
After being transferred to bigger and bigger hard drives several times during the last few years my MP3s (especially the older ones) have definitely degraded in quality. Not that it matters now, my HD crashed and I lost all 36 gigs of them except for the 4 on my iPod... When I get back home (where all my cds are) from school I intend to rip most of them to Ogg.

I still need to figure out what to do about classical though. I like having my music on my hard drive, but I really dislike listening to classical and stuff like that after it's been compressed. How affordable/practical is either SACD or DVD-A now?
     
Addicted to MacNN
Join Date: Nov 1999
Location: Madison, WI
Status: Offline
Reply With Quote
Feb 15, 2003, 07:21 PM
 
Guys, Files don't "Ware out".

Thats like saying that Apples website is hard to read because so many have loaded and read the pages that the files are wareing out.

-Owl
     
Posting Junkie
Join Date: Jun 2001
Location: Baltimore, MD
Status: Offline
Reply With Quote
Feb 15, 2003, 07:32 PM
 
Originally posted by OwlBoy:
Guys, Files don't "Ware out".

Thats like saying that Apples website is hard to read because so many have loaded and read the pages that the files are wareing out.

-Owl
Yes, but everytime you change the physical location of the data there will be some degradation of the data. There will be tiny inconsistencies between the original and the copy. If you copy the same file many times those inconsistencies will add up until, eventually, they are apparent to human perception. The reason there's a limit to the data density on a storage medium is that the smaller you get, the harder it is to safe-guard against these errors, so they are very real and are a big factor in the advancement of storage technology.
     
Clinically Insane
Join Date: Nov 1999
Status: Offline
Reply With Quote
Feb 15, 2003, 07:37 PM
 
Digital copies do not degrade through generations and copying. The very nature of digital encoding makes this impossible.

However, limitations of the media on which the data is stored can cause problems. Files can become corrupted, whether through transfer to bad media, or catastrophic crashes, and such. These sorts of things can, theoretically, affect a file over time. But that is not the same as degrading through generations.
You are in Soviet Russia. It is dark. Grue is likely to be eaten by YOU!
     
Professional Poster
Join Date: Jan 2001
Location: Cairo
Status: Offline
Reply With Quote
Feb 15, 2003, 07:39 PM
 
perhaps the files were severely fragmented?
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 15, 2003, 07:45 PM
 
Originally posted by OwlBoy:
Guys, Files don't "Ware out".

Thats like saying that Apples website is hard to read because so many have loaded and read the pages that the files are wareing out.

-Owl
You are wrong. Digital is great because it makes errors rare, not impossible. Even with the addition of error correction, there is still a finite possibility of an error occurring (the trick is to push that probability out past a couple thousand years). Just sitting there isn't much of a problem (with CD's anyway, I don't trust magnetic media for long term storage because the domains can flip just by sitting there), but what they're talking about is the copy of a copy of a copy effect. You can copy an original as many times as you like, with only an occasional problem fil. When you start copying the copy, and on down the chain, though, quality will always be lost.

IOW, yes, the recopying is probably the source of the errors in your songs. There are two strategies I can think of to counteract this effect: first, make a master backup copy ASAP on some media that degrades slower than magnetic media (i.e. DVD, CD); second, you can use one of the RAID strategies that is meant to improve data integrity instead of speed. RAID 5 is one such system, IIRC, and it only requires two discs. The other flavors require 3 HDs (odds are really slim that two drives will get the same error on the same bit, so if one HD disagrees with the other two, the two win out and the error is corrected in the third). Word of warning: trying either of these methods on partitions of a single disc is unwise because your speed with be cut by a half in the case of RAID 5, and 2/3 in the others (i.e. you're sending each but 3X from the HD).

BlackGriffen

P.S. I realize that the ultimate problem with optical media (talking really long term here) is UV damage and degredation of the media itself. I'd still bet that, sitting for the most part unused and protected from UV light, optical discs will have a longer life than magnetic.
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 15, 2003, 08:41 PM
 
Originally posted by Millennium:
Digital copies do not degrade through generations and copying. The very nature of digital encoding makes this impossible.
Wrong. Analog copies are error prone because even the smallest difference in signal voltage effects the copy. Digital copies are very hard to screw up because the difference in voltage between a one and a zero are about 5V or so. It's a big leap, but definitely not impossible. Not to mention read errors from the drive, magnetic domains that sometimes 'spontaneously' flip, and write errors, plus whatever the network throws in if it is long range. How best can I put this? Digital can ignore most noise because the votage difference between a zero and a one is greater than the typical spikes caused by noise. That does not preclude atypically large voltage spikes from causing a false one or false zero.

In short: digital = good, digital + error correction = better, but still not perfect. The closest you can get to perfect is by pushing the mean time between errors out so rediculously far (using redundancy, error checking algorithms, durable media, etc) that errors are not going to noticably accumulate in the useful lifetime of the data.

To give you an example of how digital isn't perfect, consider the lengths to which many mainframes go to achieve accuracy. One such example is the RAID methods I mentioned. Another is that some mainframes will actually have two CPUs executing the same code in parallel, and comparing the answers. If the CPUs ever disagree, the process is frozen, stepped back to before the disagreement, sent to another pair of processors, and the disagreeing procs are flagged for replacement.

BlackGriffen

P.S. I better explain the why I distrust magnetic media, and why the bits flip. The information on a hard drive is stored using little magnetic dipoles (think microsopic bar magnets). So far, so good. The problem comes from the fact that, in the presence of a magnetic field that is perpendicular to the dipoles, their state will precess (like a gyroscope, basically they'll start to rotate around the magnetic field line). This isn't really a problem inside of the HD, because the local dipoles are lined up, so they don't cause each other to precess (the distant dipoles shouldn't cause a problem, and should pretty much cancel each other out). The problem comes from Earth's magnetic field (and whatever else is floating around). The problem is reduced by magnetic shielding (I have to wonder if the manufacturers aren't skimping to cut costs with the recent drop in overall HD quality, as evidenced by dropping warranty legnths), magnetic shielding is imperfect, though, and can only reduce the amount of magnetic field the plates experience, but not eliminate it. Exactly how much the field is reduced depends on the material used, it's thickness, and the geometry of the hard drive, but a reduction in strength of about 1000 sounds reasonable. What this means is that the period at which the bits precess will be very high (guessing ~ centuries or millennia, can't give an educated guess, becuase I don't know how strong those little dipoles are). So they're precessing slowly, everything should be fine, right? Well, quantum mechanics throws a monkey wrench in the whole thing. When you read a bit, you are basically measuring the spin of a bunch of atoms whose spins like to line up. The result of said measurement will either be up or down (1 or 0) for each magnetic domain (I'm pretty sure that the bits are now about the size of a single domain). Because the state has precessed some since the previous measurement, there is a finite probability that the measurement will be the opposite of what it should be. For any single bit, this probability is miniscule, even if the media has been sitting for a long time. Do the measurement to billions and billions of bits, though, and probability will eventually catch up to you. The only way to prevent magnetic media from doing this, AFAIK, is to refresh the drive (reading the data "straightens out" the dipoles), and use redundancy, and error correction, et al. There is a problem with refreshing because you have to balance refresh rate against mechanical wear and tear. At any rate, I like optical media because it doesn't have this problem. It is also possible, if you want to go really long term, to make it from things that don't degrade fast instead of just cheap plastic. Oh, well, everything I've just discussed is primarily important for long term and very large archives. For the average user, more basic measures (i.e. optical master backup copy that is kept in a dark place) should work fine.

Edit: Reduced "though"ing, and otherwise attempted to clean up.
( Last edited by BlackGriffen; Feb 15, 2003 at 08:47 PM. )
     
Professional Poster
Join Date: Feb 2002
Location: Vallejo, Ca.
Status: Offline
Reply With Quote
Feb 15, 2003, 09:05 PM
 
CD hostage....

You say you are hearing random pauses. Are you, by any chance, using pro speakers on any of the machines where this is so?
In a realm beyond site, the sky shines gold, not blue, there the Triforce's might makes mortal dreams come true.
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 15, 2003, 09:06 PM
 
I should also add that since mp3s discard all unnecessary information, they are more vulnerable to degredation (every bit is important) than uncompressed files. I don't know, though, about lossless compression. Each bit is more important, but because the file is smaller, any one file is less likely to have errors. mp3 is certainly benefited by it's reduced size, but any changed bit has far more potential to be a catosrophic loss (possibly even invalidated the frame, depending on how the decompression algorithm works). Whereas, uncompressed is more resilient, at worst giving you popping noises. Compressed lossless should be middle of the road.

Just had an idea I'm going to send to the ogg people...

BlackGriffen
     
Mac Elite
Join Date: Jul 2001
Status: Offline
Reply With Quote
Feb 16, 2003, 12:11 AM
 
Good call, Mac Zealot, my Dad said that his sometimes futz around.

I use my excellent headphones, NOVA40 says on the side, which are so well padded you could use them as earmuffs, and give much better quality than the iMac speakers.

Given the statements in the last few posts, it looks like as long as I'm on single copies, this is bound to happen. *sigh* I just remembered something from Scientific American - a sort of Rosetta Stone for 2002 A.D. The plan was to make a billion tiny metal balls, practically indestructible (perhaps tungsten carbide?), and inscribe them with explanations of the various Roman-alphabet languages. In monospaced font, etched right in! All you'd need is a good optical microscope to read one. The balls can be embedded in books and in the walls of places of learning - someone's bound to find ONE of them.

The idea is, if civilization fails or a germ war breaks out and almost all humans are destroyd (heck, get rid of all of us, make it aliens who need to learn about us!) then the future readers will have a relatively easy time of understanding the ancient writing. Cool idea, in these apocolyptic times.

What did that have to do with my music collection? Nothing much, except to make a wonerful thought of storing data almost permanently as holes in sheets of diamond.
Actual conversation between UCLA and Stanford during a login on early Internet - U: I'm going to type an L! Did you get an L? S: I got one-one-four. L! U:Did you get the O? S: One-one-seven. U: <types G> S: The computer just crashed.
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 16, 2003, 01:50 AM
 
Originally posted by cdhostage:
Good call, Mac Zealot, my Dad said that his sometimes futz around.

I use my excellent headphones, NOVA40 says on the side, which are so well padded you could use them as earmuffs, and give much better quality than the iMac speakers.

Given the statements in the last few posts, it looks like as long as I'm on single copies, this is bound to happen. *sigh* I just remembered something from Scientific American - a sort of Rosetta Stone for 2002 A.D. The plan was to make a billion tiny metal balls, practically indestructible (perhaps tungsten carbide?), and inscribe them with explanations of the various Roman-alphabet languages. In monospaced font, etched right in! All you'd need is a good optical microscope to read one. The balls can be embedded in books and in the walls of places of learning - someone's bound to find ONE of them.

The idea is, if civilization fails or a germ war breaks out and almost all humans are destroyd (heck, get rid of all of us, make it aliens who need to learn about us!) then the future readers will have a relatively easy time of understanding the ancient writing. Cool idea, in these apocolyptic times.

What did that have to do with my music collection? Nothing much, except to make a wonerful thought of storing data almost permanently as holes in sheets of diamond.
Don't take my gloom and doom setniments about truly long term (100 years plus) storage to heart. Your problem is the "copy of a copy" syndrome. As I explained, make a master copy on optical media (from which you can have clean files to copy to any new HD) and you'll be fine, or be more careful when moving from hard drive to hard drive (actually not sure how to do this method without making it tedious, there has to be a way of copying files that uses error correction that isn't manual, isn't there?).

BlackGriffen
     
Mac Elite
Join Date: May 2001
Location: Up north
Status: Offline
Reply With Quote
Feb 16, 2003, 02:18 AM
 
I still find it hard to believe in degradation of digital files this bad...

I have programs that I have copied back and forth many, many times, over many many years.. and they don't crash. If there were degradation in the file, wouldn't one expect the program to cough out?? I mean chaning a bit in a program could be the equivalent of changing a instruction from an equal to operator, to a mov instruction. Which would be a pretty big logic error... so why don't programs stop working after being copied so many times over so many years?

And even if there were small errors, like say a 10 bits were flipped, I doubt copying the files say 10 times would be enough to propagate the errors enough to degrade your mp3's.. it seems more reasonable to me that iTunes is screwing up your files.
     
Clinically Insane
Join Date: Apr 2000
Status: Offline
Reply With Quote
Feb 16, 2003, 02:25 AM
 
Files do not "degrade".
     
Addicted to MacNN
Join Date: Sep 2001
Location: Toronto
Status: Offline
Reply With Quote
Feb 16, 2003, 02:46 AM
 
Re mp3 quality: I encode all my music in 192 VBR, using a LAME encoder. There is no audible difference to the original. If you REALLY want to be on the safe side encode everything at 256. What Hifi in he UK recently run a test on mp3 quality and concluded that that's the optimum bitrate.
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 16, 2003, 11:51 AM
 
Originally posted by 11011001:
I still find it hard to believe in degradation of digital files this bad...

I have programs that I have copied back and forth many, many times, over many many years.. and they don't crash. If there were degradation in the file, wouldn't one expect the program to cough out?? I mean chaning a bit in a program could be the equivalent of changing a instruction from an equal to operator, to a mov instruction. Which would be a pretty big logic error... so why don't programs stop working after being copied so many times over so many years?

And even if there were small errors, like say a 10 bits were flipped, I doubt copying the files say 10 times would be enough to propagate the errors enough to degrade your mp3's.. it seems more reasonable to me that iTunes is screwing up your files.
If I understand correctly, he has been transfering multiple gigabytes several times. How big was the program you were copying? I'm willing to bet about 1000 times smaller. IOW, you would have to copy it 1000 more than cdhostage has to see the same number of errors he has.

Also, it would seem that there is some rudimentary error detection in the mp3 standard. In the least, because mp3s crop frequencies above 22kHz or so, any degredation is likely to manifest as a pop or click that will add frequencies in that range. Recognizing such as a bad frame, the player refuses to play it.

Last of all, there is definitely some iTunes foul ups here, too. Nothing I have described so far could explain the exchanged ID3 tags between two songs. That is far too specific for it to be any kind of thermal or otherwise random effect.

BG

P.S. Cipher: Everything I've said has been true, although my estimate numbers like the angular frequency of the dipoles may be far enough off for the effect to be negligible except over very long periods of time or multiple terrabytes of data. In fact, if you want to get to the really nitty gritty details, it's impossible for files not to degrade over time without intervention because of thermodynamics. Cases oxydize, magnetism becomes randomized if the device is ever heated too much (not sure about the temperature of this one, could be up to 1000F before kT noticably overrides energy that keeps ferromagnets aligned, definitely depends on the material, though) or heavily jarred (you know you can demagnetize things by hitteng it with a hammer, right?). The plastic in optical media will degrade, especially if exposed to UV light. Even diamonds slowly turn into graphite at our temperature and pressure. There is also the issue of signal noise and cross talk, which can both cause spikes big enough to make a one time copy error. Since these were external drives, were the contacts clean? That's about the only factor under user control that I can remember. At any rate, Cipher, the point I've been trying to make is that digital reduces the number of errors by ignoring & straightening out small differences, it does not and cannot eliminate all errors, though. At best, with sufficient redundancy and error correction algorithms, and good equipment maintenance, you can drop the probability of error ludicrously low (low enough to be considered certainty). Such methods are prohibitively expensive in this case.
     
Admin Emeritus
Join Date: Oct 1999
Location: Zurich, Switzerland
Status: Offline
Reply With Quote
Feb 16, 2003, 01:32 PM
 
I have to disagree with the likelihood of errors. I don't think you'd encounter them from just a couple of copies, under normal circumstances.

My guess was that at some point in your chain of copies, there was a failing drive or IDE controller. Rev. A blue and white G3s have dud IDE controllers that don't get along with most modern drives, causing severe data corruption in the finished copy.

As for how hard drives work... it's not that simple. Modern drives actually use a system called PRML (or the enhanced EPRML), for "Partial Response, Maximum Likelihood", which means, roughly, that the drive is half-guessing at what it wrote. But it works, and massively increases drive capacity.

As for voltages of digital signals... the whole "0 is 0V and 1 is 5V" thing is an uber-simplification. TTL logic represents 0-2V (roughly) as a 0 and 3-5V (roughly) as a 1. (2-3V is a buffer zone.) CMOS logic represents binary as percentages of the supply voltage, which can be anywhere from 3-45V. 0-30% is a 0, 70-100% is a 1, and 30-70% is the buffer. Then there's differential signalling, which uses two wires and a ground to cancel out interference (FireWire and differential SCSI [HVD and LVD] use this). And in a lot of signalling (such as the pits and lands on a CD), it's not actually the voltage (or light) level that represents data: it's the transition (indicating a 1) or lack of a transition (indicating a 0) from high to low or low to high.

Upshot is, if you've gotten so many errors that you have problems in your MP3s, I'd start doing diagnostics on your equipment to see where they're occurring, cuz the errors won't single out your MP3s, they will affect ALL files.

tooki
     
Posting Junkie
Join Date: Jun 2001
Location: Baltimore, MD
Status: Offline
Reply With Quote
Feb 16, 2003, 06:23 PM
 
At least in my case, I was copying multiple gigabytes of data from drive to drive, occasionally between file formats, definitely between OSs, and some of those drives were higher quality that others. After 7 years of so of this, there was definite problems with some of my MP3s that hadn't been there before. Various factors other than just moving the files around undoubtedly were a factor as well such as power spikes, physical shock, and things like that.
     
Mac Elite
Join Date: Jun 2001
Location: Dundee, Scotland
Status: Offline
Reply With Quote
Feb 16, 2003, 07:48 PM
 
While digital media may "degrade" with time, digital files do not degrade by copying them.

I anything, recopying them would make them last longer - placing them on a newly written peice of media.

Filesystems fragment. Disks degrade. CDs get scratches and computer just wear out.

But the more copies you make of an MP3 file the better chance you have of having a backup.
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 16, 2003, 08:36 PM
 
Originally posted by sambeau:
While digital media may "degrade" with time, digital files do not degrade by copying them.

I anything, recopying them would make them last longer - placing them on a newly written peice of media.

Filesystems fragment. Disks degrade. CDs get scratches and computer just wear out.

But the more copies you make of an MP3 file the better chance you have of having a backup.
There's a difference, though. Every time you move a bit, there's a small probability that said bit will be flipped. The master copy won't suffer from this kind of copy error, but the copy, if a bit gets flipped, won't be quite as good as the original. If you then copy said damaged file, there's a small chance of it getting worse. I'll grant that you can make as many copies of an original as you like, only having to worry about wearing out the media (very very slow), but when you start getting in to the copy of a copy thing, the quality will degrade after enough generations.

BG
     
Clinically Insane
Join Date: Apr 2000
Status: Offline
Reply With Quote
Feb 16, 2003, 09:41 PM
 
Originally posted by BlackGriffen:
If I understand correctly, he has been transfering multiple gigabytes several times. How big was the program you were copying? I'm willing to bet about 1000 times smaller. IOW, you would have to copy it 1000 more than cdhostage has to see the same number of errors he has.

Also, it would seem that there is some rudimentary error detection in the mp3 standard. In the least, because mp3s crop frequencies above 22kHz or so, any degredation is likely to manifest as a pop or click that will add frequencies in that range. Recognizing such as a bad frame, the player refuses to play it.

Last of all, there is definitely some iTunes foul ups here, too. Nothing I have described so far could explain the exchanged ID3 tags between two songs. That is far too specific for it to be any kind of thermal or otherwise random effect.

BG

P.S. Cipher: Everything I've said has been true, although my estimate numbers like the angular frequency of the dipoles may be far enough off for the effect to be negligible except over very long periods of time or multiple terrabytes of data. In fact, if you want to get to the really nitty gritty details, it's impossible for files not to degrade over time without intervention because of thermodynamics. Cases oxydize, magnetism becomes randomized if the device is ever heated too much (not sure about the temperature of this one, could be up to 1000F before kT noticably overrides energy that keeps ferromagnets aligned, definitely depends on the material, though) or heavily jarred (you know you can demagnetize things by hitteng it with a hammer, right?). The plastic in optical media will degrade, especially if exposed to UV light. Even diamonds slowly turn into graphite at our temperature and pressure. There is also the issue of signal noise and cross talk, which can both cause spikes big enough to make a one time copy error. Since these were external drives, were the contacts clean? That's about the only factor under user control that I can remember. At any rate, Cipher, the point I've been trying to make is that digital reduces the number of errors by ignoring & straightening out small differences, it does not and cannot eliminate all errors, though. At best, with sufficient redundancy and error correction algorithms, and good equipment maintenance, you can drop the probability of error ludicrously low (low enough to be considered certainty). Such methods are prohibitively expensive in this case.
BG: Of course, but we're not talking about the effects of thermodynamics on hard drive platters or read heads... we're talking about in general.

Copying an mp3 to another hard drive, over a network, to a memory stick... the nature of digital media is that data doesn't change when copied...

If the things you mentioned were short-term issues, and if data did indeed change as much as some people in here are claiming, well, your applications would become corrupted amazingly often, copying them would be futile, etc.

It's easy to imagine a PSD having a bit or two changed, or an mp3, but think about applications and code and stuff...
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 16, 2003, 10:40 PM
 
Originally posted by Cipher13:
BG: Of course, but we're not talking about the effects of thermodynamics on hard drive platters or read heads... we're talking about in general.

Copying an mp3 to another hard drive, over a network, to a memory stick... the nature of digital media is that data doesn't change when copied...

If the things you mentioned were short-term issues, and if data did indeed change as much as some people in here are claiming, well, your applications would become corrupted amazingly often, copying them would be futile, etc.

It's easy to imagine a PSD having a bit or two changed, or an mp3, but think about applications and code and stuff...
You seem to be making the same mistake others are making. This isn't a case of:

master copy -> copy, copy, copy, copy...

Which is what you seem to be describing. This is:

original copy -> copy -> copy2 -> copy3 -> copy4 ->...

Every arrow in the process is a chance for errors to enter the picture (via interference or noise in the signal, bad signaling equipment, etc). Now, let's say (for example's sake) that every other arrow introduces an error. What you get is something like this:

master -> copy, copy*, copy, copy, copy*, copy, copy*, copy*

original -> copy* -> copy* -> copy** -> copy*** -> copy*** -> copy *** -> copy****

Where each star is indicates that the copy damaged the file somehow. The original in both cases is fine. The copies, though, are overall far worse in the second case than in the first.

Tooki has it right: I was overstating the processes I was talking about, but the fundamental point stands: digital is better, not perfect. The best way to think of it is that digital is just another error correction algorithm. In both cases, you're sacrifising theoretical information density for increased surety of tranmission. Every algorithm has its holes where it will fail to catch an error, though. In the triplicate the information case, for instance, if the same error ever appears in two files (highly unlikely, but still possible), then the error will override the correct bit. It is possible to reduce the odds further by quintuplicating the data, requiring the same error to appear in 3 files, but 5 copies is already far down the road of diminishing returns (depending on the error rate, I suppose). What I'm trying to get at is that there is a tradeoff between information density, and surety of transmission. If we could build a reliable analog computer, for instance, where the voltage level stood for a floating point value, such a computer could theoretically run far faster than the fastest digital computer ever could (for every 1 or 0 a digital computer sends, an analog computer could send an entire number). The problem with such a computer, though, is that it would be extremely susceptible to noise (even the slightest crosstalk or wiggle on the signal changes the value, compared with digital where the noise needs to be pretty big to effect the signal), and thus it's answers would be essentially completely unreliable.

BlackGriffen
     
Mac Elite
Join Date: Oct 2001
Location: Edinburgh
Status: Offline
Reply With Quote
Feb 16, 2003, 11:39 PM
 
On a similar note, I recently started to question whether copying a CD by reading the original (in an AIFF file, so not compressing it) then burning a new one would give you an identical copy, as I'd always thought it would. To have perfect copies you have to assume that any CD ROM drive you use to read the original disk will be able to extract 100% of the data from it. If this is the case, then why do different audio CD players have clear differences when it comes to the amount of detail that they extract from the music? Is it all down to the D-A conversion, amplification circuitry etc? Can the transport and pickup really make no difference?
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 16, 2003, 11:53 PM
 
Originally posted by talisker:
On a similar note, I recently started to question whether copying a CD by reading the original (in an AIFF file, so not compressing it) then burning a new one would give you an identical copy, as I'd always thought it would. To have perfect copies you have to assume that any CD ROM drive you use to read the original disk will be able to extract 100% of the data from it. If this is the case, then why do different audio CD players have clear differences when it comes to the amount of detail that they extract from the music? Is it all down to the D-A conversion, amplification circuitry etc? Can the transport and pickup really make no difference?
I'll just make a list of possible issues (in order of when they apply):
  • badly scratched disc
  • read errors (includes any skipping due to bumping)
  • accuracy of the D-A converter (includes sample rate, and quality of the power supplied to it).
  • quality of the amplifier (also linked to power supply, but includes issues of any frequency dependent signal gain and/or phase shifts that distort the sound)
  • Quality of the speakers
And the whole time the signal is going through wires that may have pickup (valid concern after D-A convesion), and said wires are connected by contacts that can introduce their own noise (mostly if they aren't clean).

BlackGriffen
     
Mac Elite
Join Date: Jul 2001
Status: Offline
Reply With Quote
Feb 17, 2003, 10:43 AM
 
There seem to be some people here who don't understand - after a copy is made, the original is destroyed.

Sometimes immediately, for defragmenting purposes (just a copy from the same drive to another spot on the same drive) sometimes it's a backup and the main drive fails, sometimes it's when I want to start over, make a clean install of 10.1.5 or 10.2.1 instead of having a whole lotta garp left over from installed and discarded stuff.

See? The heads on a HD are not perfect - there can be any number of problems that can show up to make one bit be recorded (or read/transferred to the next device) as the wrong value. A little extra electricity coming through the power cord, a subtle magnetic pulsing from my headphones, perhaps something built right into MP3 format so that they don't last (play and alter ). I don't know!
Actual conversation between UCLA and Stanford during a login on early Internet - U: I'm going to type an L! Did you get an L? S: I got one-one-four. L! U:Did you get the O? S: One-one-seven. U: <types G> S: The computer just crashed.
     
Mac Elite
Join Date: Oct 2001
Location: Edinburgh
Status: Offline
Reply With Quote
Feb 17, 2003, 04:17 PM
 
Originally posted by BlackGriffen:
I'll just make a list of possible issues (in order of when they apply):
  • badly scratched disc
  • read errors (includes any skipping due to bumping)
  • accuracy of the D-A converter (includes sample rate, and quality of the power supplied to it).
  • quality of the amplifier (also linked to power supply, but includes issues of any frequency dependent signal gain and/or phase shifts that distort the sound)
  • Quality of the speakers
And the whole time the signal is going through wires that may have pickup (valid concern after D-A convesion), and said wires are connected by contacts that can introduce their own noise (mostly if they aren't clean).

BlackGriffen
Thanks BG, but most of these apply to what happens to the digital signal once it is read from the disc. What I'm wondering about is purely that first stage, the reading of data from the disc, before any D-A conversion, amplification etc takes place. What I can't get my head round is this: If you can use a CD ROM on a computer to install software, for example, then that implies to me that it can read 100% of the data on the CD correctly. As most CD ROM drives are pretty crappy flimsy things, that would mean that virtually any old CD player is capable of reading 100% of the data on a CD, apart from those that are too scratched to play at all, of course. If that is true, then why do hifi manufacturers place so much emphasis on the quality of their laser pick ups, clamping mechanisms etc? I can see how the subsequent D-A conversion and amplification could make a big difference to the detail in the sound, but what advantage is there in trying to improve the actual data-reading part, if any CD player can pick up all the data?

If my assumption that all CD players / CD ROMs can read 100% of the data is wrong, then how on earth are you able to install software from them?
     
Professional Poster
Join Date: Jul 2001
Location: Dis
Status: Offline
Reply With Quote
Feb 17, 2003, 05:29 PM
 
Originally posted by talisker:
Thanks BG, but most of these apply to what happens to the digital signal once it is read from the disc. What I'm wondering about is purely that first stage, the reading of data from the disc, before any D-A conversion, amplification etc takes place. What I can't get my head round is this: If you can use a CD ROM on a computer to install software, for example, then that implies to me that it can read 100% of the data on the CD correctly. As most CD ROM drives are pretty crappy flimsy things, that would mean that virtually any old CD player is capable of reading 100% of the data on a CD, apart from those that are too scratched to play at all, of course. If that is true, then why do hifi manufacturers place so much emphasis on the quality of their laser pick ups, clamping mechanisms etc? I can see how the subsequent D-A conversion and amplification could make a big difference to the detail in the sound, but what advantage is there in trying to improve the actual data-reading part, if any CD player can pick up all the data?

If my assumption that all CD players / CD ROMs can read 100% of the data is wrong, then how on earth are you able to install software from them?
The assumption may be wrong, sort of. The only thing I can think of is that a data disk has enough error correction algorithms to compensate, where an audio CD does not. Either that, or they're touting a useless feature (wouldn't be the first time).

BG
     
Mac Elite
Join Date: Oct 2001
Location: Edinburgh
Status: Offline
Reply With Quote
Feb 17, 2003, 07:02 PM
 
Originally posted by BlackGriffen:
The assumption may be wrong, sort of. The only thing I can think of is that a data disk has enough error correction algorithms to compensate, where an audio CD does not. Either that, or they're touting a useless feature (wouldn't be the first time).

BG
OK, after doing a bit of investigation I've found out that error correction is indeed the answer. Data CDs do indeed have more rigorous error checking on them. However what's interesting is how audio CD error checking works - if the disc isnt being read properly it will obviously try and correct it, and if it fails to read the missing data properly it "bridges" the gap by interpolating what should be there (unless it's really missed a lot, in which case it jumps or whatever). So what happens is that the fine detail in the music is reduced. If your CD ROM drive is doing this when you read a CD before copying, the copy will obviously not be as good as the original, even though everything has been digital throughout. The difference will only be apparent when you play it on a better CD player of course, but still it knocks the "perfect digital copy" argument on the head.
     
Mac Enthusiast
Join Date: Oct 2001
Location: Edmonton, AB
Status: Offline
Reply With Quote
Feb 17, 2003, 11:45 PM
 
I have 3 gigs of MP3's, all of which have been transferred to and from CD's at least once. Some are a year old. I've NEVER noticed quality problems. Try encoding at 256 kbps.
     
Clinically Insane
Join Date: Apr 2000
Status: Offline
Reply With Quote
Feb 18, 2003, 12:30 AM
 
Well, of course if you make a million copies of something a bit might flip here and there, but people are overstating this problem, making it ridiculous...
     
   
Thread Tools
Forum Links
Forum Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On
Top
Privacy Policy
All times are GMT -4. The time now is 06:49 AM.
All contents of these forums © 1995-2015 MacNN. All rights reserved.
Branding + Design: www.gesamtbild.com
vBulletin v.3.8.8 © 2000-2015, Jelsoft Enterprises Ltd., Content Relevant URLs by vBSEO 3.3.2