Artificial Intelligence Meets the Beatles

//

In recent weeks, we’ve seen stories about how robots are threatening to replace newspaper writers and encyclopedia authors. It looks like music critics might be next.

Researchers at Lawrence Technological University have used artificial intelligence technology to analyze the style of a particular band, pinpoint nuances in the evolution of the band’s sound and then accurately map the group’s musical development over time.

Such an algorithm could be useful for searching, browsing and organizing large music databases and it can also work to more closely match a listener’s musical preferences.

Music Not For Everyone — Really

Specifically, the scientists have focused on the work of The Beatles, analyzing songs from each of the 13 Beatles’ studio albums that were released in the U.K. The system works by transposing the audio of a song into a spectrogram — a visual representation of sound. The spectrogram is then broken down into thousands of numerical descriptors indicating shapes, patterns and distribution of similar pixels.

Here’s where it gets interesting. By using pattern recognition and statistical algorithms, the A.I. program was able to sequence all 13 albums in the proper order. The A.I. not only figured out that “Please, Please Me” was the Beatles first record, it also correctly placed subsequent albums — “With the Beatles,” “Beatles for Sale,” “A Hard Day’s Night” — in the proper order, all the way up to “Let It Be” and “Abbey Road.”

In essence, the program did a kind of massive compare-and-contrast analysis, figuring out which songs were most alike, and which were least alike. The A.I. even figured that the songs on “Let It Be,” the last album commercially released by the band, were in fact recorded before the songs on “Abbey Road,” the band’s second-to-last record.

DNews: Do Animals Like Music?

“People who are not Beatles fans normally can’t tell that ‘Help!’ was recorded before ‘Rubber Soul,’ but the algorithm can,” says lead researcher Lior Shamir. “This experiment demonstrates that artificial intelligence can identify the changes and progression in musical styles by ‘listening’ to popular music albums in a completely new way.”

The research team has run similar experiments on recordings by Queen, U2 and Abba.

And this is interesting: In one of the early beta experiments, the team tried to run the Shania Twain catalog through the system, but the display simply generated the number “666″ on infinite loop, triggered a blackout in southeastern Canada, and shorted out three dozen orbital satellites. We kid, we kid.

via Science Daily

Credit: Wikimedia Commons