How an artificial intelligence can become a truly musical companion

Music has long been hailed as one of the last frontiers of human creativity; and music is not just composition, but also, and equally important, expressive performance. It is the performers and the musicians who bring a piece to life, producing an expressive rendition that brings out dramatic, affective, and emotional qualities which may engage and deeply affect listeners; this effect cannot be replicated by a machine – until now. Gerhard Widmer is an award-winning computer scientist, professor and head of the Institute of Computational Perception at Johannes Kepler University Linz. As part of their long-term research on AI & music, his team developed a computer accompanist that permits expressive co-performance between a human and a machine, based on learned computational models of expressivity. It is arguably the first demonstration of a machine playing together with a human in a truly musical way. At Falling Walls, Gerhard will talk about what makes music come alive, what it means for a computer to learn fundamental principles of expressive (piano) performance, and how, by using this learned knowledge, it can contribute to expressive music making – not as a replacement for human pianists, but as a musical companion.

Gerhard Widmer was voted Science Breakthrough of Year in the category Art and Science at the Falling Walls Science Summit. To find out more go to

The Falling Walls Science Summit was part of Berlin Science Week 2021 and hosted by the Falling Walls Foundation.