iTunes Nabs a Counterfeiter

Subject: iTunes Nabs a Counterfeiter
From: JohnInKansas
Date: 18 Jun 07 - 08:38 PM

A current Newsweek article relates some details of the late career of "The greatest pianist almost noone ever heard of," Joyce Hatto. She was a fairly good, but not exceptional, pianist and received mediocre reviews during a brief early concert carerr, and then retired to private teaching. Near the end of her life, a series of "mastered" CDs were released under private label that received phenomenal reviews. A total of 119 CD releases appeared, to rave reviews.

A chance discovery, when someone with an iTunes account plugged one of her CDs in, led to a forensic analysis that revealed that apparently all of the Joyce Hatto CD performances were simply copied from the performances of other great pianists.

To oversimplify how iTunes indicated the fraud, early recordings had no identifying information tacked onto the recordings, so iTunes (Gracenote) uses the "duration of the piece" to identify individual performances from a library, and identified a piece on the Joyce Hatto CD:

When New York financial analyst Brian Ventura loaded Hatto's version of Liszt's Etudes into his computer, the online Gracenote database recognized it as Hungarian Laszlo Simon's 1987 recording of the same piece, based on the fact that both recordings were exactly the same length—down to one seventy-fifth of a second.

The current Newsweek article, June 25, 2007 issue, is at Grand Theft Piano for those who'd like more info.

A "related article" that appeared in an earlier issue, gives some of the Tech: How the Hatto Hoax Was Revealed

This earlier article was also posted here previously at:

I Read It In The Newspaper (individual post).

So if you're gonna steal someone else's performance for your own CD, leave the last three seconds off the piece – or add a fart at the end – I guess.

John

Subject: RE: iTunes Nabs a Counterfeiter
From: Skivee
Date: 18 Jun 07 - 08:59 PM

wow, that really sucks.

Subject: RE: iTunes Nabs a Counterfeiter
From: Zhenya
Date: 18 Jun 07 - 09:37 PM

I'd heard the Hatto story a while back, and this really did seem to be fraudulent.

However, I had my own odd experience along these lines last year. I went to add a Scottish CD to iTunes, and the tracks came up as being another Scottish CD, by a different performer. I owned that one too, so I compared the track listings. Although the songs and tunes were all different, the two CDs had the same number of tracks, and the track lengths for each track were almost identical, going through the entire CD.

I had both of these CDs around for years, both by reasonably well-known Scottish performers (I have to confess I don't remember exactly the 2 performers now - I believe one was the group Ossian, but it's been awhile since this happened.) I have no reason to think either of these CDs was fraudulent. I think this was just one really weird coincidence.

Subject: RE: iTunes Nabs a Counterfeiter
From: JohnInKansas
Date: 18 Jun 07 - 09:50 PM

Like the DNA database that the Government wants, the coincidences can occur (identical twins have identical DNA) but in this case additional forensic research confirmed the fraud.

Music released on CDs, at least by major producers, generally has embedded/encoded track data for identification of the source, but the old tracks used in the Hatto releases probably came from vinyl or tape, where there was no such information.

For the old releases, apparently a very precise track length is used, but for a newer one where there's additional info, it may be somewhat looser and used only if an indy release omitted the encoded identification(?).

We may have an expert who can explain it all to us, perhaps.

The effort/ability to "identify" individual tunes in the absence of standard release data certainly explains why all the "online sellers/renters" insist on constantly pushing their "latest versions" with new and improved spyware usage tracking built in.

John

Subject: RE: iTunes Nabs a Counterfeiter
From: M.Ted
Date: 19 Jun 07 - 02:13 AM

According to the article, the recordings had actually be electronically altered to conceal their source. It strikes me that it would have been a fairly simple thing to hit the time dilation button--if he had, he might have gotten away with it--

Subject: RE: iTunes Nabs a Counterfeiter
From: JohnInKansas
Date: 19 Jun 07 - 03:04 AM

It should be suspected that the Newsweek article oversimplifies things.

The release of 119 CDs obviously would include a large number of individual tracks. All that really was required is that someone recognize that "this track" sounds a lot like "that other performer," and be sufficiently curious to compare the two closely.

That the iTunes registry found a "correct" name for one of the tracks, and that the one who noticed was able to push for further investigation, may well have been pretty much coincidental.

The "right listener" might well have simply said "that sounds like my favorite performance by x" to trigger the eventual verification that the CDs were faked.

But it makes a better story to claim that "the new technology found it." The new technology may have raised the odds slightly, but it still was left to the person with the suspicion to seek a real investigation to prove what was going on.

John

Subject: RE: iTunes Nabs a Counterfeiter
From: Geoff the Duck
Date: 19 Jun 07 - 05:39 AM

A few weeks back I was looking at the Gracenotes CD Database and other alternative sites which perform a similar task.
When I searched in Wikipedia it gave info about how Gracenotes works and it is a pretty simple analysis based only on the number and length (time) of the tracks on a CD. It assumes that no two CDs will contain the same number of tracks with matching lengths in the same order. Obviously it is possible, as in the case of the two Scottish CDs, for two CDs to have matching track lengths and therefore give an incorrect identification. BLICKY
Other databases may use a different method of differentiation.
Further reading here http://en.wikipedia.org/wiki/Acoustic_fingerprint

Quack!
GtD.

Subject: RE: iTunes Nabs a Counterfeiter
From: The Fooles Troupe
Date: 19 Jun 07 - 06:59 AM

"assumes that no two CDs will contain the same number of tracks with matching lengths in the same order. Obviously it is possible, as in the case of the two Scottish CDs, for two CDs to have matching track lengths and therefore give an incorrect identification"

It's just like hash tables in computing - more than one input can give the same hash code, and there is only only space in a hash code table to have one item, so you have to have a way to record and identify other inputs that match the same hash code.

Obviously Gracenotes was designed by someone who had did not understand 'the appropriate field of research'... :-)

Theory DOES have its uses.... :-P

Subject: RE: iTunes Nabs a Counterfeiter
From: M.Ted
Date: 19 Jun 07 - 11:51 AM

It can be very difficult to distinguish between two performances of the same piece--depending a lot on the piece, of course. Some pieces leave a lot of room for interpretation, and others don't. For the pieces that don't, about the only way to distinguish one artist from another is to listen for mistakes.

The identification of the piece in question hinged on the fact that there was an out-of-tune key in the older recording that was there in Ms. Hatto's recording, as well.