Do computers leave behind a digital fingerprint on files?

[esldude]

Dennis

They were sent to one of your correct addresses. The files will almost certainly have expired by now.

 

Alex,

 

It was sent to my discontinued address which is why I didn't see it. I do still have some access to the inbox. I am downloading the files now. Will let you know what I think.

[sandyk]

I have you down @bham. Is that correct ? If so, it should have gone there ,not the old address.

[esldude]

I have you down @bham. Is that correct ? If so, it should have gone there ,not the old address.

 

That is the old address. That is where I found them. I PM'd you my new address just now.

 

Alex these two files have a nearly 3 db difference in loudness. No surprise they will sound different. They also are equalized differently up until about 5 khz. And there is a small speed difference. I doubt they are the same master.

 

Just out of curiosity I tried them using Jud's proposal. Mono them and put one in each channel. The level difference and different playback speed morphs the soundstage interestingly over the course of playback. Early in the track when they are timed fairly close the EQ differences are noticeable as an effect as well.

[Audio_ELF]

So I'm watching Scorpion last night and yes I know it's TV so likely it's all science fiction. In the show, they mentioned that in the field of computer forensics they catch criminals based upon what's left behind. Supposedly each computer leaves behind traces even though the file looks the same. This made me think of how in the computer audiophile community there is so much talk over noise that's somehow transmitted while still being bit perfect. I know that tags can me modified and the file still sound the same so it doesn't sound so this concept doesn't sound so completely unrealistic. If there is truth to this concept in computer forensics, it could explain what some hear even though two files look the same. Does anyone here know if this is all made for tv or if there really are "digital fingerprints" on files?

Just to get back on topic ... when working forensically they also have to go through processes which can be replicated by any future defence investigators. For that reason when forensically examining a computer the first thing they do is duplicate the HDD - this is done at a low level so that any deleted data is also replicated. In forensic examinations very rarely would the technician work on an original.

[sandyk]

That is the old address. That is where I found them. I PM'd you my new address just now.

 

Alex these two files have a nearly 3 db difference in loudness. No surprise they will sound different. They also are equalized differently up until about 5 khz. And there is a small speed difference. I doubt they are the same master.

 

Just out of curiosity I tried them using Jud's proposal. Mono them and put one in each channel. The level difference and different playback speed morphs the soundstage interestingly over the course of playback. Early in the track when they are timed fairly close the EQ differences are noticeable as an effect as well.

 

Dennis

I wasn't concerned about the level difference. To my ears, and that of another member, the 24/96 version is far more detailed, with even his voice sounding way better. Neither does the 16/44.1 version have anywhere near the same soundstage. Through my system there is simply no comparison, and the level difference has nothing to do with it.

 

Alex

[firedog]

I've skimmed through this one, and I wonder if some of you actually read what you write.

This has to be (one of) the most wheels off threads in the history of the forum....

[Richard Dale]

CMO = Closed-Minded Objectivist

 

That's a new one on me. I am in the 'subjectivist' camp normally - I love my single ended EL84 valve amps and Nordost speaker cables etc. But I am also a computer programmer and know that the computers I use wouldn't work if files got corrupted when they were copied. When we are discussing the sound of cables, or how they can't sound different, we are referring to effects in the 'real world' which is an open system that we certainly don't fully understand. On the other hand, what goes on inside a computer at the level of the 'file' abstraction is a closed system. It is 100% defined by humans and has nothing to do with the physical world - it is an abstract concept running on a computer. So saying that copying files adds 'invisible artifacts' is equivalent to saying that unexpected hidden things happen when you add the integers 2 + 2 in maths and don't quite get 4 as a result. It doesn't work like that is all I can say.

[kumakuma]

That's a new one on me. I am in the 'subjectivist' camp normally - I love my single ended EL84 valve amps and Nordost speaker cables etc. But I am also a computer programmer and know that the computers I use wouldn't work if files got corrupted when they were copied. When we are discussing the sound of cables, or how they can't sound different, we are referring to effects in the 'real world' which is an open system that we certainly don't fully understand. On the other hand, what goes on inside a computer at the level of the 'file' abstraction is a closed system. It is 100% defined by humans and has nothing to do with the physical world - it is an abstract concept running on a computer. So saying that copying files adds 'invisible artifacts' is equivalent to saying that unexpected hidden things happen when you add the integers 2 + 2 in maths and don't quite get 4 as a result. It doesn't work like that is all I can say.

 

I am in complete agreement with you.

 

In regards to the phrase, it is an expression that Alex uses from time to time to refer to those of us who question his claims that bit identical CD rips done with better power supplies can sound different.

[kumakuma]

Dennis

I wasn't concerned about the level difference. To my ears, and that of another member, the 24/96 version is far more detailed, with even his voice sounding way better. Neither does the 16/44.1 version have anywhere near the same soundstage. Through my system there is simply no comparison, and the level difference has nothing to do with it.

 

Alex

 

I would have a hard time attributing the improvement to higher resolution when the levels and equalization are also different.

 

Do you hear the same dramatic improvement when you listened to Mario's test files?

 

To me this is a better test of the benefits of higher resolution as there is only difference between the files, not three.

[esldude]

Dennis

I wasn't concerned about the level difference. To my ears, and that of another member, the 24/96 version is far more detailed, with even his voice sounding way better. Neither does the 16/44.1 version have anywhere near the same soundstage. Through my system there is simply no comparison, and the level difference has nothing to do with it.

 

Alex

 

Okay Alex. This nothing to do with resolution. It is about the mastering. Were both 44 or both 96 the same one would sound better. So I am at a loss seeing the point.

[blownsi]

The way I see it...

 

If a bit is a bit then the following sound the same:

 

PC, Mac, Sonos & Squeezebox transports

Windows, OSX, IOS, Android & Linux

Itunes, MPD, Audirvana, Foobar, J River & Amarra

Switching & Linear power supplies

USB, Ethernet, Coax & Optical cables

All SPDIF Convertors

CD, DAT, NAS, SSD & SD Cards

Quantity of RAM - 4gb, 8gb or 16gb

Ram Disk, Spinning Drive & Solid State

Fan & Fanless

20 or 120 Background processes running

 

 

If you believe that any of the above is untrue then you are part of the audiophile conspirators not against them.

[Jud]

That is the old address. That is where I found them. I PM'd you my new address just now.

 

Alex these two files have a nearly 3 db difference in loudness. No surprise they will sound different. They also are equalized differently up until about 5 khz. And there is a small speed difference. I doubt they are the same master.

 

Just out of curiosity I tried them using Jud's proposal. Mono them and put one in each channel. The level difference and different playback speed morphs the soundstage interestingly over the course of playback. Early in the track when they are timed fairly close the EQ differences are noticeable as an effect as well.

 

Would you say it felt easier to identify the files as different when they were combined into one track as left and right channel, when they were separate tracks, or about the same?

 

Edit: Did/would you try it equalizing loudness?

[sandyk]

The way I see it...

 

If a bit is a bit then the following sound the same:

 

PC, Mac, Sonos & Squeezebox transports

Windows, OSX, IOS, Android & Linux

Itunes, MPD, Audirvana, Foobar, J River & Amarra

Switching & Linear power supplies

USB, Ethernet, Coax & Optical cables

All SPDIF Convertors

CD, DAT, NAS, SSD & SD Cards

Quantity of RAM - 4gb, 8gb or 16gb

Ram Disk, Spinning Drive & Solid State

Fan & Fanless

20 or 120 Background processes running

 

 

If you believe that any of the above is untrue then you are part of the audiophile conspirators not against them.

 

An interesting take on the subject. I wonder what percentage of C.A. readers honestly believe that this is true with all of the above ?

[sandyk]

DELETED.

A simple QUICK REPLY should not have resulted in 2 copies !

[sandyk]

Okay Alex. This nothing to do with resolution. It is about the mastering. Were both 44 or both 96 the same one would sound better. So I am at a loss seeing the point.

 

Sorry, but I disagree . Given good equipment, it is a good example of the benefits of high res over 16/44.1 despite the obvious differences in level etc. One person even remarked that with the high res version you can even hear the spit in his throat, or words to that effect.

All the technical trickery (quackery?) that you are using will never reveal why many agree that high res sounds better than 16/44.1.

Playing silly buggers with file manipulation, inverting etc. will only result in degradation and a loss of low level precision.

[mansr]

The way I see it...

 

If a bit is a bit then the following sound the same:

 

PC, Mac, Sonos & Squeezebox transports

Windows, OSX, IOS, Android & Linux

 

With bit-perfect output there's no difference. Out of the box, some of those will do nasty things.

 

Itunes, MPD, Audirvana, Foobar, J River & Amarra

 

Again, no difference in bit-perfect mode. Any filters they apply might of course sound different.

 

Switching & Linear power supplies

 

For analogue components they can make a difference. Not so for computers or other purely digital devices.

 

USB, Ethernet, Coax & Optical cables

 

No difference provided they meet relevant specs. A bad cable can obviously cause data corruption.

 

All SPDIF Convertors

 

SPDIF to/from what?

 

CD, DAT, NAS, SSD & SD Cards

 

No difference.

 

Quantity of RAM - 4gb, 8gb or 16gb

 

No difference once there's enough. 4GB is more than plenty.

 

Ram Disk, Spinning Drive & Solid State

 

No difference.

 

Fan & Fanless

 

Apart from mechanical noise from the fan, no difference.

 

20 or 120 Background processes running

 

Provided the audio buffers never under-run, no difference. Too high load can starve the audio player of CPU time leading to drop-outs.

[esldude]

Just to get back on topic ... when working forensically they also have to go through processes which can be replicated by any future defence investigators. For that reason when forensically examining a computer the first thing they do is duplicate the HDD - this is done at a low level so that any deleted data is also replicated. In forensic examinations very rarely would the technician work on an original.

 

How dare you jump into a thread like this and get it back on topic? I tell you some people have no courtesy whatsoever. After several of us have invested time and effort to derail a topic and you try and just throw all that away! Really!!!!

 

:)

[rn701]

Does every thread on this forum end in this exact discussion/argument regardless of the original topic?

 

Anyway, on file systems such as NTFS, each file has a set of parameters such a security ID (SID) for the owner, an access control list (ACL) and other stuff. It is stored by the file system separate from the actual file data.

 

Using certain utilities, this info can be "copied" (recreated) on the target file system when a file is moved or copied. It may or may not be useful, if for example the file is copied to another domain, which would not know the SID's etc. from the source domain.

 

These parameters could, I suppose, be matched to the source domain/file system. If copied to a different type of OS/file system the info is generally lost because it would be meaningless.

 

This is overly simplified but close enough for tv scifi/cop shows.

[mansr]

Does every thread on this forum end in this exact discussion/argument regardless of the original topic?

 

Whenever sandyk joins, yes.

[esldude]

Would you say it felt easier to identify the files as different when they were combined into one track as left and right channel, when they were separate tracks, or about the same?

 

Edit: Did/would you try it equalizing loudness?

 

Well it isn't really possible to equalize loudness for the sake of comparison. I looked a bit more in detail today. There are EQ differences that are 6-8 db hotter in the 44 version around 8 khz. They grow gradually from around 5 khz. Then from 10 khz upward the 44 version is at or more than 15 db hotter. Of course there isn't as much energy in the upper regions so the lower band still sets perceived loudness for the most part. My manner of looking at it may also show not just EQ. It could be a different compression method that leaves more upper band artifacts than another. Looking at blown up details of the waveform I believe the percussion tracks are out of phase in the two versions. No way of knowing which are correct.

 

Now my initial listen without changing either file it was immediately apparent imaging was over to one side which was due to the basic loudness difference. It was apparent the EQ was bloating the apparent size in some frequencies in one direction. Then the speed difference made them out of synch after a few seconds.

 

I would say in this instance the files were obviously different. I would say the mono comparison made it more apparent more quickly the differences were large and multi-faceted. So this may not be a good file to judge by though it is an interesting experience to hear.

[sandyk]

Whenever sandyk joins, yes.

 

In case you didn't notice, I was drawn into the thread, as is so often the case , by wgscott in reply 9 , or the creepy stalker Kumakuma .

I am not about to let arrogant know-it-alls like you bring my name into these threads without being permitted to reply.

If you don't want to see me needlessly involved in this type of thread then don't drag me into it.

Simples !

[Tony Lauck]

Copies of files may contain the same samples and yet be different. One explanation is the tag information. Even if there is no tag information, a WAV file has control fields in addition to the audio samples and these can differ. This means that playing two files with identical audio samples will invoke a difference sequence of I/O operations and processor instructions and this alone may result in a different sounding playback due to cross talk from digital noise into the DAC and other analog equipment.

 

Using WAV files I have converted them to FLAC and back to WAV. The audio samples are identical in the two WAV files, but often the file contents (as proven by an MD5 checksum) are not. It is possible that Cookie listened to two such files on a system that was susceptible to these differences. With different conversion software and or a different playback system she might have heard no differences. But note that two identical file copies of the same file may sound different and they will be different. They will have different file names and will be stored on different disk locations. It is not possible to argue logically that the two files can not sound different. Instead, one has to work with an open mind and investigate exactly what is going on.

 

Cookie didn't do this. She's not a computer engineer or scientist. She is an exceptional recording engineer. If she heard differences I'd be pretty confident that there were differences since this is her business. But as to the explanation, she is not qualified to say when there might or might not be a difference. Her approach is to avoid unnecessary production operations that may create problems. If she can sell WAV files for a higher price then the FLAC files produced from the WAV, more power to her. (I have always saved the money and purchased the FLAC when the master was PCM.)

 

I can hear differences between real time playback of WAV and FLAC files. I can not hear differences between original WAV files and WAV files that have been passed through FLAC and converted back to WAV. If I could then I would be motivated to investigate this further. In the past, I have encountered people who claimed to have heard these differences, but they would never follow through with careful experiments to use their hearing as an instrument to isolate the source of differences they heard.

[Jud]

In the past, I have encountered people who claimed to have heard these differences, but they would never follow through with careful experiments to use their hearing as an instrument to isolate the source of differences they heard.

 

Most of the proposed experiments I have seen have had a good chance of not showing differences.

 

What I have read that has interested me is Cookie's mention that she has trained people to hear these differences. I would be very interested in reading more about that, if it would be of value when not actually with Cookie and listening to the differences as she described them.

[mansr]

Copies of files may contain the same samples and yet be different. One explanation is the tag information. Even if there is no tag information, a WAV file has control fields in addition to the audio samples and these can differ. This means that playing two files with identical audio samples will invoke a difference sequence of I/O operations and processor instructions and this alone may result in a different sounding playback due to cross talk from digital noise into the DAC and other analog equipment.

 

It is extremely unlikely that such differences result in measurable, let alone audible, difference at the output of a DAC.

[wgscott]

Does every thread on this forum end in this exact discussion/argument regardless of the original topic?

 

Anyway, on file systems such as NTFS, each file has a set of parameters such a security ID (SID) for the owner, an access control list (ACL) and other stuff. It is stored by the file system separate from the actual file data.

 

Using certain utilities, this info can be "copied" (recreated) on the target file system when a file is moved or copied. It may or may not be useful, if for example the file is copied to another domain, which would not know the SID's etc. from the source domain.

 

These parameters could, I suppose, be matched to the source domain/file system. If copied to a different type of OS/file system the info is generally lost because it would be meaningless.

 

This is overly simplified but close enough for tv scifi/cop shows.

 

Extended attributes on OS X are encoded in the resource fork, which would be copied with the file but does not change the checksum, which reports the bit content of the data fork.