20 February 2007

Hats off

by Matt Rubinstein at 3:13 am

Joyce_Hatto_2.jpgA number of celebrated recordings by the late British pianist Joyce Hatto have just been “uncovered” as fakes. Hatto performed extensively in the 1950s and 1960s but retired from the stage in 1976 after being diagnosed with cancer. She apparently dedicated the rest of her life to a series of brilliant recordings of just about every difficult solo piano piece ever, which were released by her husband’s label, Concert Artist. From her 2006 obituary in The Times:

The discography she assembled latterly, surveying almost completely the piano works of Haydn, Mozart, Beethoven, Schubert, Schumann and Brahms, is a trusty Baedeker of the standard repertoire. But composers outside the Austro-German canon were not neglected: there is, naturally, much Chopin and Liszt, but also Scarlatti, Debussy, Prokofiev and her father’s beloved Rachmaninov. Moreover, this outstanding conspectus was capped by the summa of piano technique, Godowsky’s reworkings of Chopin’s études, tackled in their entirety by only the most fearless.

Everybody seems to have loved these recordings. But it turns out that at least some of them bear surprising similarities to tracks and even whole albums recorded by other talented 20th-century pianists. Questions about Hatto’s unlikely prolificacy had been circling for a while, but it all came down when an anonymous critic for Gramophone magazine stuck a Hatto CD into his computer and was told that it was really a recording by Lászlo Simon previously released by BIS Records.

Now, anyone with iTunes and an Internet connection will recognise something like the miraculous Gracenote® or CDDB database, which has enhanced our ripping experience considerably since the 1990s. CDs typically don’t contain any digital information about artists or titles—they don’t know what they are, and so copying one to your computer always involved typing in all of these details so you could later identify your music. CDDB set up an online database allowing users who had completed this tedious task to upload the information so that nobody else would have to.

The database’s cleverest part was its ability to identify every CD shoved into somebody’s computer. I’d always thought that CDs had some identifying number, equivalent to an ISBN, that was encoded somewhere in the CD. It turns out they can, but a lot of them don’t. So CDDB creates a more-or-less unique ID for a CD by crunching its table of contents (the list of where on the disc each track starts) through some kind of algorithm. The result is that if an album is reissued by another label but the table of contents isn’t disturbed, CDDB will still identify it correctly. Or if you rebadged someone else’s CD under your own name, CDDB would call you on it.

Gramophone magazine sent some Hatto recordings, including the Liszt, to audio expert Andrew Rose to check them out. Interestingly, Rose says that only ten of Lászlo Simon’s twelve Liszt études had made it onto Hatto’s version, and that the fourth track was shrunk by 0.02%, or about a tenth of a second—a pretty wacky result, but not enough to throw out CDDB’s calculations. But the other two tracks on Hatto’s version aren’t Simon’s: one of them is Minoru Nojima and is eight seconds shorter than the Simon, and the last track remains unidentified.

You can see the problem: Gracenote® or any other database that works the same way wouldn’t match Hatto’s Liszt to Simon’s, as the Gramophone story goes: they’re not the same CD, so the fingerprint would be all different. If the database didn’t already hold Hatto’s version, it would return a blank. That doesn’t mean that the Hatto tracks aren’t copies—there is a lot of analysis that suggests that they are—but it does mean that the “discovery” story is probably a fiction, or was at least misreported—it may well have been another CD that was an exact copy. As usual, the <a href="http://groups.google.com/group/rec.music.classical.recordings/search?group=rec.music try this.classical.recordings&q=hatto&qt_g=Search+this+group”>newsgroups are all over it, with some suggesting that some shadowy figure had filled in the CDDB entry for Hatto’s disc with Simon’s details, and it all becomes very complicated. It’s the biggest scandal to hit the world of classical music since—I don’t know—Bond!

2 Responses to “Hats off”

  1. Alastair Says:

    Well, it could have worked like this.

    Someone gets the Hatto recording, rips them to their computer. That person recognises them as actually Simon recordings, changes the track description on their computer, and *uploads* the revised descriptions to CDDB.

    [Insert some handwaving about the ability to upload to Gracenote’s database. If not Gracenote then FreeDB.]

    A second person, one with access to the Gramophone editors, also gets the Hatto CD, and downloads the revised descriptions.

    Not saying that this is a very likely scenario, just that it could have happened this way.

    Another explanation is that the anonymous whistleblower was fully aware that the tracks were Simon recordings at the outset, and used the computer-based identification as a way of adding credibility (at least amongst the non-mattrubinstein.com.au-reading public). It’s the “computer says no” ploy, in other words.

    A great story anyway, look forward to seeing how it plays out.

  2. Matt Rubinstein Says:

    That’s true Al—I think that’s more or less how the newsgroup conspiracy theorists have it. I guess it’s also possible that the CDDB algorithm has developed a certain fuzziness: so that, I don’t know, if someone reissues an album with a bonus track the database will ID the album and fill in the tracks it does have, that might be a useful enhancement. But who knows?

Leave a Reply