Karaoke Scene's Karaoke Forums
http://karaokescene.com/forums/

Digital Watermarking (new topic)
http://karaokescene.com/forums/viewtopic.php?f=26&t=36878		 Page 1 of 1
Author:  chrisavis [ Mon Feb 13, 2017 6:33 am ]
Post subject:  Digital Watermarking (new topic)
c. staley wrote:
Paradigm Karaoke wrote:
c. staley wrote:
Let's look at both sides of this (deflection) issue:
#1. If there is a watermark, so what? Does it make a difference? Does it somehow prevent Chris from jumping on a torrent
(which he's obviously very familiar with) and sharing it with his friends?

no, it would back up your claim, that's all. exactly what you ask everyone else to do.
someone claims it, back it up. and to me that is not an unreasonable expectation from anyone.

And how exactly, would something like this be "backed up?" Share computer source code or markers or a roadmap that detail the process or how to even detect it? As much as Chris would like that, it's not going to happen. Tell me what you think is reasonable that doesn't compromise anything. All Chris has done is proclaimed that because he hasn't found anything, there can't be anything there.

It's nothing more than a stalemate.


It absolutely is about not finding anything because there is nothing to find.

Whether I used a free file analyzer or a commercial one doesn't make any difference. They both do the same thing.

More Detail.....

All computer files are comprised of 0's and 1's. There is no way to modify a file by adding a watermark without changing those 0's and 1's. (one could use ADS but there is no ADS in use with Chip's files). Chip claims that the Red Peter's tracks are uniquely identifiable to the purchaser which means every track will have a unique watermark comprised of a unique set of 0's and 1's. The files I used in my test were all identical so it isn't possible for them to have unique ID's.

Chip could easily prove the watermarking claim without divulging any code or secrets.

If the file has a watermark identifying the purchaser, then take a sample of files and name who purchased them. I will happily volunteer my files as well as a those from a few other KJ's I know that have bought some Red Peters tracks.

I am sure Chip will have some sort of excuse for not doing that as well.
Author:  JimHarrington [ Mon Feb 13, 2017 7:00 am ]
Post subject:  Re: Digital Watermarking (new topic)
chrisavis wrote:
(one could use ADS but there is no ADS in use with Chip's files)


Plus, non-Windows systems wouldn't preserve the ADS data anyway.

chrisavis wrote:
Chip claims that the Red Peter's tracks are uniquely identifiable to the purchaser which means every track will have a unique watermark comprised of a unique set of 0's and 1's. The files I used in my test were all identical so it isn't possible for them to have unique ID's.

Chip could easily prove the watermarking claim without divulging any code or secrets.


All that's really required is to run the MD5 checksum on two copies of the same track purchased by two different people. If they have the same checksum, they are identical on the bit level and do not contain a watermark. If they differ, they might include a watermark. (Or they might not.) In fact, it wouldn't be necessary to make extraneous copies of the files. All that would be required would be for two or more people who have the same track to run it through an MD5 checksum calculator (http://www.winmd5.com/) and post the result. Since MD5 is a one-way hash, no sensitive information would be shared.
Author:  chrisavis [ Mon Feb 13, 2017 7:26 am ]
Post subject:  Re: Digital Watermarking (new topic)
Also, let me be clear..... I am not questioning whether the tracks are watermarked at all. I am questioning Chip's claim that they are uniquely watermarked in a manner that the purchaser can be identified from a watermark.
Author:  c. staley [ Mon Feb 13, 2017 9:48 am ]
Post subject:  Re: Digital Watermarking (new topic)
JimHarrington wrote:
All that's really required is to run the MD5 checksum on two copies of the same track purchased by two different people. If they have the same checksum, they are identical on the bit level and do not contain a watermark. If they differ, they might include a watermark. (Or they might not.) In fact, it wouldn't be necessary to make extraneous copies of the files. All that would be required would be for two or more people who have the same track to run it through an MD5 checksum calculator (http://www.winmd5.com/) and post the result. Since MD5 is a one-way hash, no sensitive information would be shared.
Sorry, but to use your line "I'm not at liberty to correct your misinformation."

How long have you been a programmer?

Please read the first 2 paragraphs:
https://en.wikipedia.org/wiki/MD5

Have a nice day.
Author:  chrisavis [ Mon Feb 13, 2017 10:48 am ]
Post subject:  Re: Digital Watermarking (new topic)
c. staley wrote:
JimHarrington wrote:
All that's really required is to run the MD5 checksum on two copies of the same track purchased by two different people. If they have the same checksum, they are identical on the bit level and do not contain a watermark. If they differ, they might include a watermark. (Or they might not.) In fact, it wouldn't be necessary to make extraneous copies of the files. All that would be required would be for two or more people who have the same track to run it through an MD5 checksum calculator (http://www.winmd5.com/) and post the result. Since MD5 is a one-way hash, no sensitive information would be shared.
Sorry, but to use your line "I'm not at liberty to correct your misinformation."

How long have you been a programmer?

Please read the first 2 paragraphs:
https://en.wikipedia.org/wiki/MD5

Have a nice day.



Nothing in that article pokes any holes in the ability of MD5 to generate hashes against files and for those hashes to be used for digital fingerprinting.

In other words, the MD5 hash generated from a Red Peters track I purchased should be different from the MD5 hash generated from a file someone else purchases *provided* they are uniquely watermarked (which they are not).

Try again......
Author:  JimHarrington [ Mon Feb 13, 2017 1:29 pm ]
Post subject:  Re: Digital Watermarking (new topic)
c. staley wrote:
JimHarrington wrote:
All that's really required is to run the MD5 checksum on two copies of the same track purchased by two different people. If they have the same checksum, they are identical on the bit level and do not contain a watermark. If they differ, they might include a watermark. (Or they might not.) In fact, it wouldn't be necessary to make extraneous copies of the files. All that would be required would be for two or more people who have the same track to run it through an MD5 checksum calculator (http://www.winmd5.com/) and post the result. Since MD5 is a one-way hash, no sensitive information would be shared.
Sorry, but to use your line "I'm not at liberty to correct your misinformation."


You aren't "at liberty" to correct what I said because what I said was entirely correct.

c. staley wrote:
How long have you been a programmer?


Professionally? About 16 years. At all, about 35 years. I started programming in 1982 when my father bought an IBM PC Model 5150. For the last 16 years I've been the principal developer on a series of software packages that business forms and promotional products companies use to put their inventories on the web. I'm a fully qualified LAMP full-stack developer and server administrator. When I get tired of being a lawyer, I will probably program full-time.

c. staley wrote:
Please read the first 2 paragraphs:
https://en.wikipedia.org/wiki/MD5

Have a nice day.


Nothing in the first two paragraphs of that article (or anywhere else) contradicts what I said. In fact, the following sentence confirms what I said:

It can still be used as a checksum to verify data integrity, but only against unintentional corruption.

In order for two copies of a track to be digitally watermarked in a useful way, they must be able to be differentiated in some manner. As Chris explained, that means there will be some sort of change in the sequence of 1s and 0s that make up the file. When you want to verify that the copy of a file you received is identical to the source (i.e., hasn't been corrupted), you can compare the MD5 checksum of the file you have to the known MD5 checksum of the source file. If they match, you know that there are no differences between the source and the copy.

If two people each have a "watermarked" copy of a given track, the MD5 checksums for their respective files WILL differ. If those people produce the same checksum for their copies, we know that the two files are not distinguishable and therefore cannot be individually watermarked. (This would not be the end of the inquiry unless the MD5 checksums are the same, because it's possible for them to differ yet not be individually watermarked--but I suspect the inquiry will end there.)

Now, it's true that it's possible to corrupt a file intentionally in a way that doesn't change the MD5 checksum, but surely you're not suggesting that someone is out there intentionally corrupting your tracks to beat the test, right? I mean, what would be the point of that?
Author:  c. staley [ Mon Feb 13, 2017 6:20 pm ]
Post subject:  Re: Digital Watermarking (new topic)
You were expecting some sort of confirmation or denial from me?

Really?
Author:  Karaoke Croaker [ Mon Feb 13, 2017 6:43 pm ]
Post subject:  Re: Digital Watermarking (new topic)
if any change in a file causes it to not be an identical copy then media shifting isn't using illegal counterfeit copies if you just lower the volume of the track or change the font because then it's not an exact copy any longer. If you were to run your MD5 tester thingy they wouldn't come up with the same results with any minor change to the file.....like removing the brand name or logos or changing the volume? Very interesting!
Author:  jdmeister [ Tue Feb 14, 2017 11:28 am ]
Post subject:  Re: Digital Watermarking (new topic)
Quote:
I started programming in 1982 when my father bought an IBM PC Model 5150. For the last 16 years I've been the principal developer on a series of software packages that business forms and promotional products companies use to put their inventories on the web. I'm a fully qualified LAMP full-stack developer and server administrator.


Interesting.. I teach Unix/Linux programming/administration at the college level, and am a published author in several Linux and Unix themed magazines as well a bit of MD5 and LAMP stack on 2600 magazine. I started my classes on BSD Unix administration back in 1997..

(Windows networking and hardware diagnostics also Chris)
It's always fun reading these threads.. :mrgreen:
Author:  chrisavis [ Tue Feb 14, 2017 11:49 am ]
Post subject:  Re: Digital Watermarking (new topic)
c. staley wrote:
You were expecting some sort of confirmation or denial from me?

Really?


Confirmation has already been established.
Author:  JimHarrington [ Tue Feb 14, 2017 12:22 pm ]
Post subject:  Re: Digital Watermarking (new topic)
Karaoke Croaker wrote:
if any change in a file causes it to not be an identical copy then media shifting isn't using illegal counterfeit copies if you just lower the volume of the track or change the font because then it's not an exact copy any longer. If you were to run your MD5 tester thingy they wouldn't come up with the same results with any minor change to the file.....like removing the brand name or logos or changing the volume? Very interesting!


"Identical copy" is not the legal definition of "counterfeit."
Page 1 of 1 All times are UTC - 8 hours
Powered by phpBB® Forum Software © phpBB Group
https://www.phpbb.com/