Few symphonies are as famous as Beethoven’s Ninth Symphony. This statement is supported by the fact that it is undoubtedly playing in your head even as you read this.
Few symphonies are less well known – at least among Beethoven’s – than his Tenth Symphony, primarily because Beethoven never completed the work.
However, after his death in 1827, drafts and sketches for the Tenth Symphony were discovered.
Since then, lovers and connoisseurs of the great composer’s work have been wondering what Beethoven’s Tenth might have become.
“In 1988, musicologist Barry Cooper took the risk of finishing the first and second parts of the piece,” writes Ahmed Elgammal, director of the Rutgers University’s Arts and Artificial Intelligence Laboratory, in The Conversation.
“He brought together 250 bars of music from Beethoven’s sketches to create what he thought might be the first part of the Symphony, as Beethoven intended. The paucity of the sketches left after Beethoven did not even allow us to imagine what the next parts could become. “
However, as the 250th anniversary of Beethoven approached, the era of artificial intelligence had already arrived. For Matthias Röder, director of the Karajan Institute of Salzburg, it was an irresistible temptation to combine the great composer and the promising technology.
Elgammal and Röder became part of a team that came together to accomplish a daunting task: to develop a form of machine learning that could help complete Beethoven’s Tenth Symphony. Together with them worked composer Walter Verzova (known for writing a signature jingle for microprocessor manufacturer Intel), computer music expert Mark Gotham and musicologist-pianist Robert Levin, who “previously completed a series of unfinished works by Mozart and Johann Sebastian Bach.”
Funding for the project was provided by Deutsche Telekom.