The word "gesture" typically refers to a physical movement of the body to convey meaning or expression. As used in musicological discourse by theorists and composers, "gesture" more commonly refers—admittedly ambiguously—to an aspect of musical sound that evokes a metaphorical sense of physicality and movement.
An understanding of musical gesture is important to human musical expression. For a computer to be a successful improviser of music, it must be able to recognize and produce expressive musical gestures. How can a computer, which has no sentient body and no experience of physicality, understand the concept of musical "gesture"?
This research project focuses on the perception and cognition of gesture: not physical bodily movements, but the metaphorical evocation of gesture in musical sound. The question is whether the cognition and characterization of musical gesture can be algorithmically described in such a way that a computer can be programmed to use it effectively in the processes of composition and improvisation. The goal is to devise ways that a computer can “understand”—characterize and categorize—the gestural nature of music and can use that information to generate its own original musical materials.
I described an approach to computer characterization of musical gesture at the 3rd International Conference on Music and Gesture in Montréal in 2010. I explained the Gestural software, "A Method for Computer Characterization of 'Gesture' in Musical Improvisation", in the proceedings of the the International Computer Music Conference in Lubljana, Slovenia in 2012.
There have been many performances of in-concert improvisations using this software, and the work has been presented at conferences, research centers, and universities worldwide. An additional discussion of the Gestural system and its extensions, "Expressive Gesture : A technique for the use of gesture descriptors in algorithmic improvisation" appears in the proceedings of the Journées d'Informatique Musicale in Bayonne, France in 2019.
This page is written and maintained by Christopher Dobrian.
This page was last modified October 3, 2020.