Tech: How the Hatto Hoax Was Revealed

The inside story of the digital sleuthing that exposed Britain's greatest pianist as a fraud.

WEB EXCLUSIVE
By Brian Braiker
Newsweek

May 20, 2007 - By the time Joyce Hatto died of ovarian cancer at age 77, she had released 119 albums. Beethoven, Mozart, Chopin—every one of them had been mastered. She was called the greatest pianist that no one had ever heard of. Fame finally came posthumously, but not for her virtuosity at the ivories. In February it was discovered that each of her brilliant records was very likely to have been plagiarized. Put on a Joyce Hatto CD and you'll really hear Yefim Bronfman, John O'Conor, Vladimir Ashkenazy and dozens more, but not Joyce Hatto. Fittingly, the Hatto myth unraveled much in the same way it was created: with a little enterprising digital tinkering.
When New York financial analyst Brian Ventura loaded Hatto's alleged version of Liszt's Etudes into his computer, the iTunes' Gracenote database recognized it as Laszlo Simon's 1987 recording of the same piece. But how? In 1980, when Philips and Sony established the standard technical specifications for audio compact discs (which were subsequently adopted by the international body governing such specifications), there was no need to include anything other than music on CDs. There was no call for, say, encoding song titles and artists names into the disc because back in 1980 nobody was playing CDs on their laptops. These same standards still govern audio CD manufacturing today.

So what did Gracenote see that had eluded the expertly trained ears of critics who had showered Hatto's recordings with nothing but praise? Every CD is composed of tiny chunks of audio, 1/75th of a second each, according to Ty Roberts, Gracenote cofounder and chief technology officer. When a CD is placed into a stereo or a computer, all the player can tell is how many tracks live on the disc and where they begin and end. Say track 2 of a CD starts at 3 minutes and 30 seconds into the CD. By the time the song begins, 15,750 blocks of audio, each 1/75th of a second in duration, have ticked by. (Do the math yourself: three and a half minutes is 210 seconds. Multiply that by 75 and you get 15,750 of those mini chunks of audio, called frames.). By the time you get to the 10th track, hundreds of thousands, often more than a million, of frames have ticked by. These are the numbers Gracenote reads.

"It's like we're creating a phone number for the CD," says Roberts. When you stick a CD into your laptop and iTunes tells you that it's checking in with Gracenote, what it's doing is reading how many tracks and frames are on your disc and searching the 6-million-strong CD database for a match. "If you only have three or five tracks, it's very hard for someone else to have a recording that is exactly the same length to 75th of a second. The chance you have 13 tracks that have exactly the same starting position is something like 10 to the negative 13th power." So when Gracenote told Ventura that he had loaded a Laszlo Simon disc, it was very likely something fishy was afoot. "It's a little murder-mystery thing: a shoe print in the show," says Roberts. "The shoe print doesn't fit the woman."

Ventura e-mailed the composer and pianist Jed Distler, who had reviewed numerous Hatto discs. Distler then e-mailed Gramophone magazine editor James Inverne, who in turn commissioned France-based sonic expert Andrew Rose on Valentine's Day to analyze the Hatto recording—just three days before the magazine's February deadline. The alleged Hatto tracks were uploaded onto the Internet for Rose, who bought the Laszlo Simon recording at eMusic. He opened both tracks on his computer with Adobe Audition, editing software that displays sound as waves on his monitor. "The program allows me to position the tracks to play simultaneously. You can see the wave patterns," says Rose. He could tell Hatto was a fraud before he even pressed play. "I recognized immediately they were rather too similar to be recordings by different people. We had a match."

What Rose discovered next was that someone—most likely Hatto's husband and producer William Barrington-Coupe, who runs the Concert Artist label she recorded for—had digitally manipulated the tonal quality of the track to make it sound different from the original. The culprit boosted the bass and reduced the treble, altering the timbre of the piano. Other songs were suspicious, as well. Another track on the CD was clearly not Laszlo Simon's playing. But was it Hatto's? Rose went online to hunt down matches.

He used the 30-second sound clips shoppers hear at Amazon.com and on iTunes to find a matching Liszt performance. He found one that was close, but not quite exact, on a disc by Japanese pianist Minoru Nojima, released in 1993. Played simultaneously, the tracks sound quite similar, but then gradually slide out of synch. The one attributed to Hatto is played impossibly, almost blindingly, fast—no wonder critics raved. The waveforms on Rose's monitor, however, looked undeniably similar; Nojima's was just slightly longer. So Rose compressed it to see if it would line up perfectly with Hatto's. Naturally, it did. Someone had gone to the trouble of digitally speeding up the Nojima recording—while managing to maintain the piano's original pitch—in order to hide the origin of the "Hatto" disc. "I am really astonished at how much they had been able to get away with without making it immediately apparent," he says.

Further investigating revealed plagiarism, of varying degrees of sophistication, on other discs, as well. Gramophone had its story a hair before deadline. The British media, always in love with a good hoax, leapt on board. More plagiarized discs were uncovered by passionate amateurs. ("It became a great parlor game: Spot the Joyce Hatto," says Rose.) After initial denials, Barrington-Coupe ultimately made a confession of sorts, which appeared first on the Gramophone Web site. What remains unknown is the extent to which Hatto was aware that she was part of a great swindle. To Rose, given the amount of sophisticated digital manipulation involved, it seems "inconceivable" that she didn't know. "All the evidence points to her being part and parcel of the whole thing: the interviews that she gave, the letters she wrote. I can't see how she wasn't aware of what was going on. The question is really was anyone [other than Barrington-Coupe] ever involved in handling the technical aspects of the things?" We may never know—Hatto certainly isn't talking.

Indeed, she may have been "fortunate," in Rose's words, to die before her unmasking. But her story does have one last excellent ironic twist. Large portions of Gramophone had already gone to press by the time the hoax story had finally been written. Included in the pages that had been irrevocably printed in the issue that exposed Joyce Hatto: a glowing review of a Joyce Hatto CD.