iTunes Uncovers Musical Hoax

holy_calamity writes "The reliance by iTunes on the CDDB has burst open a musical fraud in the usually staid world of classical piano. Albums by the much vaunted British pianist Joyce Hatto, who died in June 2006, are identified by the iTunes player as belonging to other performers. A more scientific analysis by an audio remastering firm has found that none of Hatto's works appear to be hers. Her husband, who produced all her albums, says he 'cannot explain' the similarities."

Re:Acronyms

[gQuigs]

And you didn't correct it for the masses...

Compact Disc Database
http://en.wikipedia.org/wiki/CDDB [wikipedia.org]

Re:Acronyms

[Thuktun]

Hint: In summaries, you should generally state what an acronym stands for, unless it is well understood by the vast majority of the intended audience. When in doubt, spell it out.

Or provide a link. [wikipedia.org] This is hypertext, for pity's sake.

Re:live performances?

[fistfullast33l]

From the unsourced Wikipedia article [wikipedia.org]:

She stopped playing in public in the 1970s, having never attained much prominence as an artist. The retired critic James Methuen-Campbell heard two of her recitals in London's Wigmore Hall and recalls a pianist with an efficient and careful technique, but with an inability to convey the overall conception of a major work. Her approach, in his opinion, concentrated on detail.

Re:Why iTunes?

[Anonymous Coward]

Um, the way I kind of understood the article is that iTunes did, in fact, "catch" it. Or at least help to identify what was going on. So in this case, it IS specific to iTunes. However, CDDB, as anyone reading /. should know is not just used by iTunes. It just happens that it was in this situation.

Re:Why iTunes?

[Hijacked Public]

The impression I took from the article is that there was strong suspicion that her CDs were fakes but no one could determine exactly which recordings from other artists had been used. iTunes, by way of CDDB, pointed the guy from Gramophone in the right direction.

So no, not iTunes directly, but since it is the Windows of music management applications it was in the right place at the right time. Also recall that these are music people and we are geeks. We may know all about CDDB and music players and which bit of software performs which task, but most normals don't know or care. Even if you try to explain it to them they will stare off in the distance, blankly, wishing they were listening to a modified version of Nojima being passed off as Hatto playing Liszt.

Re:Why iTunes?

[dschuetz]

iTunes didn't catch it, CDDB did.

Actually, neither iTunes nor CDDB caught it. The person who put the CD in caught it, when he realized that the data CDDB/iTunes returned wasn't for the CD he'd put in, but was close enough in content that he was intrigued enough to do an a/b comparison.

I'm betting a bunch of other people saw the same thing, and either didn't correct it, or said "huh" and just "corrected" the artist's name based on what they thought it was supposed to be, assuming the data in CDDB was wrong.

So kudos to the guy who noticed!

Where have I seen this before?

[Anonymous Coward]

This should have been the quote from her husband:

"It makes me laugh," he said. "The part I don't understand, the dude is trying to act like I went to his house and took it from his computer. I don't know him from a can of paint. I'm 15 years deep. That's how you attack a king? You attack moi? Come on, man. You got to come correct. You the laughing stock. People are like, 'You can't be serious.' "

Free CDDB

[Doc Ruby]

The CDDB was coded as a free repository of CD metadata. Collected by thousands of people around the Net on a worldwide, ongoing basis, by giving away the client SW which many programmers embedded into PC/Mac music players. So millions of people were prompted every time they put in an unknown CD to spend a few seconds typing in artist and song names. In exchange (though no input was required), they got most of their CDs labeled without any effort, after the CDDB was filled.

This kind of read/write database population collaboration is now well known, both in blogs and in more sophisticated databases like Wikipedia. But in the late 1990s it was revolutionary.

Then the CDDB server owners sold out to Gracenote. Gracenote required a login to access the data, which login they supplied only to licensed users. Gracenote first tried to sell CD players integrated with the CDDB, but then found more success in licensing access to iTunes and other online music distributors.

But neither Gracenote nor the CDDB programmers had produced the profitable data. The people who had were locked out. So some new programmers made a new version with the identical API and DB structure, the FreeDB [freedb.org], then datamined the CDDB to populate it. The FreeDB and its contents are GPL, so they cannot be "taken proprietary" (stolen) again. The data is free again, as is the life of this pioneering colalborative project.

If you are generating music metadata, consider submitting it to the FreeDB [freedb.org]. And try to use the FreeDB, rather than the privateer CDDB, to support you applications. And send money to the FreeDB operators whenever you can, especially if you use it.

BBC radio4 has a streaming interview

[RandomWordGenerator]

Streaming interview : Mark Lawson interviews a journalist from Gramaphone magazine (one of Joyce Hatto's champions) and talks about the issue in general, with semi-amusing lack of tech-spertise. http://www.bbc.co.uk/radio/aod/networks/radio4/aod .shtml?radio4/frontrow_mon# [bbc.co.uk]

So look at MusicBrainz

[BenFranske]

Which is a great reason to look at MusicBrainz [musicbrainz.org].

Re:Why iTunes?

[irc.goatse.cx troll]

as far as I remember, CDDB goes only by track lengths. Works some of the times, but is really a crapshoot (hence genre splitting to lower overlap).

It doesn't do any real music analysis like Musicbrainz('audio checksums') or even Pandora(manualy defined audio qualities)

Re:Blind music critics?

[David Greene]

It's fairly easy to tell the differences between two performances of the same piece. Is the pedaling the same? How about dynamics? Tempo? How is rubato used? Are the accents played similarly?

I'm by no means an expert on classical music (traditional jazz is more my thing) but even I can tell the difference between someone who is technically proficient but lacks interpretation and a real pro who understands how the whole piece fits together.

My guess is that the copied performances were not all that well known. Even if a performer is well-known, it's not necessarily the case that all of his or her performances are of such quality that they are universally recognized.

Re:Uh, what?

[SydShamino]

The first CD discovered and compared only had very limited time shifts - as little as 0.2% - and seems to have been close enough to not confuse the CDDB.

After the first CD, the investigators went out looking for older recordings to match. In these they found some with shifts as high as 15%. CDDB was no longer being used as an analysis tool; they were directly comparing the sampled data from the two sources, adjusting the time shifts until the measured aligned.

The other group that found a match (linked on page 5 or so of the article) has a musical comparison technique that they claim is based on relative timing, so they could compare without first making any adjustments.

No, really *WHY* iTunes?

[JacksBrokenCode]

Neither the Gramophone article [gramophone.co.uk] written by the critic who noticed the oddity, nor the Pristine Classical detailed write-up [pristineclassical.com] ever mention iTunes or any other specific player.

Several days ago, another Gramophone critic decided to listen to a Hatto Liszt CD, of the 12 Transcendental Studies. He put the disc into his computer to listen, and something awfully strange happened. His computer's player identified the disc as, yes, the Liszts, but not a Hatto recording. Instead, his display suggested that the disc was one on BIS Records, by the pianist Lászlo Simon. Mystified, our critic checked his Hatto disc against the actual Simon recording, and to his amazement they sounded exactly the same.

In then went a recording of Hatto playing two Rachmaninov Piano Concertos and, sure enough, his computer's CD player listed it as another - by Yefim Bronfman, conducted by Esa-Pekka Salonen, on Sony. Again, the critic compared, and again he could hear no difference.

Gramophone then sent the Hatto and the Simon Liszt recordings to an audio expert, Pristine Audio's Andrew Rose, who scientifically checked the soundwaves of each recording. They matched. "Without a shadow of a doubt," reported Rose, "10 of the tracks on the Liszt disc are identical to those on the Simon." Of the remaining two, he now feels that he has identified a further one - which he identified as being, again "without a shadow of a doubt" from a CD entitled "Nojima Plays Liszt", a 1993 release from Reference Recordings. Furthermore, his partner - who is based elsewhere with his own equipment - agrees.

If any independent research was done that shows the critic used iTunes then I have no problem, but New Scientist doesn't indicate that they did anything other than read the Gramophone and Pristine articles. Where the hell did they suddenly get iTunes?

More info on this from the horse's mouth

[naeim]

Yesterday I received an email from someone who used to be a researcher at CCRMA at Stanford about this:

---
You may find interest in following the discovery of a possible large-scale hoax in classical music.

I have been analyzing the performances of Chopin Mazurkas (http://mazurka.org.uk) and have been noticing an unusual occurence: the performances of the same two pianists always matched whenever I do an analysis for a particular mazurka. In fact, they matched as well as two different re-releases of the same original recording.

We were keeping the identity confidential due to strict libel laws in the UK and slowly building up a case. One CD set being a match could possibly be an innocent mistake, and if the record label lost business due to insinuations related to our findings... However, the story broke this past Thursday afternoon on the Gramophone website:

http://www.gramophone.co.uk/newsMainTemplate.asp?s toryID=2759&newssectionID=1 [gramophone.co.uk]

Last week, a music critic of Gramophone put a CD of Joyce Hatto's performance of the Liszt Transcedental Etudes into his CD-rom drive. The iTunes program then informed him that the pieces on the CD were correct, but the performer was different. He had that other CD and listened to both and could tell that the sounded very similar to each other. He then found using iTunes another match with Joyce Hatto playing Rachmaninov piano concertos, and again he had the original CD and could not tell a difference between them. He sent them to Pristine Audio to be analyzed by Andrew Rose, who confirmed the matches:

http://www.pristineclassical.com/HattoHoax.html [pristineclassical.com]

Andrew subsequently discovered that the Hatto performances of the Godowsky Chopin Etude Studies were also from a previously released commercial CD (although recent reports indicate that some of the tracks on the CD set are by an additional performer Marc Hamelin).

The day after the initial disclosure on the Gramophone website, CHARM (http://www.charm.rhul.ac.uk) released their findings which it had been collecting on the similarities of the Chopin mazurkas, since there was no longer any legal concerns related to releasing our corroborating findings.

http://www.charm.rhul.ac.uk/content/contact/hatto_ cover.html [rhul.ac.uk]
http://www.charm.rhul.ac.uk/content/contact/hatto_ article.html [rhul.ac.uk]

It is interesting to note that the Mazurka performances of Joyce Hatto could not be identified by the CDDB method used by iTunes to uncover the first two matches found by the Gramophone critic. The ordering of the mazurkas had been changed on the CDs, and the mazurkas were allocated differently on the two discs so that the track counts did not match. In addition, each track was timestreched by differing amounts. In the three mazurkas that I have examined in detail, the time stretching was -0.7%, -2.8%, and +1.2%. The fact that different amounts of time stretching was applied to the separate tracks leads to juicy circumstantial conclusions. It is interesting to note that Andrew Rose discovered that the Godowsky Studies had been slowed down by an incredible 15%.

Six of Joyce Hatto's CDs have been identified as copies of existing commercial recordings (as of Sunday night): three by Gramophone/Pristine Audio; one by CHARM; one by arec.music.classical.recording contributor 12 hours after the Gramophone news (so his claim to have know earlier is most likely correct); and 1 additional matching on Sunday for a source to the Chopin Etude CD set.

Hatto's mostly complete Concert Artists discography and a list of the currently identified original sources are available on her entry in

Re:No, really *WHY* iTunes?

[tmalone]

From the New York Times article:
"When he put the Hatto CD of the Liszt études into his computer, Mr. Inverne recounted, "his iTunes player identified the disc as, yes, the Liszts, but not a Hatto recording." Instead, it identified Mr. Simon as the performer."
http://www.nytimes.com/2007/02/17/arts/music/17hat t.html?_r=1&oref=slogin [nytimes.com]

I don't know if any of the articles linked in the submission mentioned it or not (I guess they didn't). My guess is that the submitter read it in the news paper, then searched online and found an article that was very similar, then submitted it without actually reading the article they were submitting.

Re:Internet phenomenon

[DF5JT]

"got any links to the posts?"

No, but I've got keywords for you:

Hatto, Deacon, Watkins, Lemken, Köhler

Search within rec.music.classical.recordings

Re:Why iTunes?

[Pollardito]

as far as I remember, CDDB goes only by track lengths. Works some of the times, but is really a crapshoot (hence genre splitting to lower overlap).

It doesn't do any real music analysis like Musicbrainz('audio checksums') or even Pandora(manualy defined audio qualities)

there was an earlier article on Slashdot [slashdot.org] that said otherwise. from that article [wired.com] :

Although the idea of using track times to identify discs existed before CDDB's time, the real trick is in using that information to find discs quickly and accurately. For some twisted reason, each time a new batch of CDs is manufactured, the factory makes a new master that often has timings slightly different from previous batches of the same disc. (I think there is a Linkin Park disc in our database with over 1,000 different TOCs.) .

It's a black art, and involves layers of hashing, fuzzy logic and other matching methods to ensure quality results. This is what Gracenote has mastered, and is far more important than just the lone idea of using track times to identify discs.

As far as acoustic recognition, Gracenote has two types of audio recognition. The simpler one is used for identifying audio files, and helps audio software catalog your music collection. The other, heavier method is very tolerant of background noise. .

The MySpace story shows one of the many ways we use this technology. One of the coolest applications is the ability to identify a song over a cell phone. We're also starting to identify music used in old TV shows, so that the rights holders/artists can be paid back royalties, as well as monitoring live radio/TV broadcasts.

Re:Why iTunes?

[Indy11]

CDDB only went by track lengths and the total length of the disc. Gracenote has evolved the CDDB recognition service to also do acoustical fingerprinting which is probably why these tracks showed up as somebody else's.

Re:TWAIN

[totally bogus dude]

I heard it as "Technology Without An Interesting Name", but I also heard that they're all just backronyms, and it was really from Kipling's "...and never the twain shall meet", referring to the disparity between interface designers and driver programmers.

Wikipedia [wikipedia.org] agrees with me, but also confirms Toolkit is the most correct expansion (although Important instead of Interesting).

Re:They may be ....

[Zonk (troll)]

No. They're true metal. Very complex layered melodies with jazz-like timing. Lyrically all of their albums are concept albums and tell a story. From their WikiPedia entry [wikipedia.org]:

Mastodon is a Grammy-nominated heavy metal band from Atlanta, Georgia. They are usually classified as a heavy metal group, but there have been debates on what specific sub-genre they belong to. There is also a clear hardcore influence on the band, unsurprising as several of its members previously played in hardcore bands.

Mastodon's style includes heavy (and sometimes quite complex and technical) guitar riffs, complex, jazz-influenced drumming, odd time signatures, and long, melodic instrumental interludes, which are all common aspects of the progressive rock genre.