French conductor Guillaume Bernie and the Nexus Orchestra used artificial intelligence to complete Beethoven’s 10th Symphony. The result is a four-minute clip, which the authors called BeethovANN Symphony 10.1.
“This is not a typo,â€� the conductor said when announcing BeethovANN Symphony 10.1 for the first time to the audience. Guillaume Bernie explains that ANN stands for “artificial neural network”, which completed “Beethoven’s next symphony”, essentially without human intervention.
“Even we don’t know what it will look like,â€� Bernie admitted to Agence France-Presse before the concert in Lausanne.
Ludwig van Beethoven began working on his 10th symphony, when he was finishing the famous ninth and ode “To Joy”, but as a result, after his death, only a few handwritten sketches and plans remained.
Florian Colombo, a computer scientist at EPFL Technical University, developed a special algorithm for its completion using deep learning. This is a subset of machine learning, where algorithms are built and function similar to machine learning, but there are many layers of these algorithms, each of which provides a different interpretation of the data it conveys. Such a network of algorithms is called artificial neural networks. In simple terms, it resembles the neural connections that exist in the human brain.
The final score was created and printed just a few hours before the performance.
To create something that could pass as a snippet from Beethoven’s 10th play, Columbo first downloaded all 16 of the master’s string quartets into his computer. He explained that chamber pieces provide a very clear picture of Beethoven’s harmonic and melodic structures.
He then asked INS to create a piece based on one of the fragments of a theme, which, according to experts, could be a draft for a new symphony.