Friday, September 21, 2007

Fantasia for Piano and Data Base


A weird story about has to do with Joyce Hatto. She was a classical pianist of the old school, something of an invalid in later years, who, through recordings, caught a buzz on the web. Bad news was that many of her multifarious recordings were discovered to be of the school plagiarism. Like grab a tape and put your name on it.

Her hubby, as described in a New Yorker story, came from something of a Wild West, though classical background. In the 50s and 60s he was involved with Bargain Bin Classical Labels that were not above usurping and dubbing the work of Euro Classicos some of who-o might have come from the Iron Curtain Side-o. Put it out, damn the royalties, laugh and chortle. [Sounds rather like some Brits approach to Blues music!]

Hatto’s unmasking was odd. Let’s look on the matter as described in the New Yorker…. One fan unwraps a Hatto disk and places it in his computers secondary storage media. His consternation redoubles as he realizes that his discovery will dampen the buzz baout Hatto.

“Ventura unwrapped the “Studies” disk .. He placed it in his computer’s disk drive and, through Apple’s iTunes software, connected to Gracenote, an Internet database of CDs. According to Gracenote, which identifies a CD by the durations of its individual tracks, Ventura had loaded the “Transcendental Studies,” but the pianist was a Hungarian named László Simon. László Simon? Ventura weighed the possibilities: Gracenote might be mistaken (mislabellings had been known to occur), or someone named László Simon had recorded the same music with precisely the same track timings. When he listened during his commute the next morning, he felt “from the very first piece, it was a remarkable recording.” On his office computer, he went to Amazon.com and found a listing for Simon’s record, including one-minute clips for most of the tracks.

“I started listening,” he recalled. “Going back and forth between the iPod and the Amazon clips for individualistic things—sudden changes in dynamics or ornaments, or a cadenzalike passage where the performer has more leeway in the interpretation. In slower pieces, it’s easier to hear subtleties. I was ninety-five per cent certain that most of the tracks were the same. So I didn’t know what to do. If she was the one that was copying, part of me didn’t want it to come out….”


It was really strange to me the first time my Internet connected computer went out and got some track name and data for a piece I had on my machine. How about you? What is the method by which this happens? People say there is a bit data base out there in Etherland that can identify a song by it’s unique length. But length cannot be that unique, even if we get down to microseconds. There has to be correlated function of some kind, right? Here is what I found on the site of Gracenote describing there patent for this technology…

A method as recited in claim 1, wherein the length information for the records in the database and for the selected recording includes information indicating the length of each segment individually, and wherein said determining includes comparing the length of corresponding segments in the selected recording and the recording corresponding to each approximately matching record.

Artificial Intelligence is about! Someday all your clothes will have RFID tags. And you will be ID’d as you walk about based on your typical unque clothes combination as garnered by a sensor network in a public building, maybe. But, as Hatto shows, the best lies have the best legs! His master’s voice is beyond duplication!

Related
Gracenotes Patents Notes
New Yorker story on Hatto

No comments: