Difference between revisions of "Computer listens to robot music"

From Crucible Network Research Projects
Jump to navigationJump to search
(Created page with "Description here Team: * Leo Impett * Rich Turner * Jim Woodhouse * Alan Blackwell Category:Projects")
 
 
Line 1: Line 1:
Description here
+
The violin family of musical instruments produces its rich effects through an astonishingly complex combination of mechanical and acoustical phenomena, many beyond our current ability to model. It seems likely that the evolution of these instruments, and the cultural and emotional effects they have on listeners, result from this very complexity.
 +
 
 +
In this project, we are exploring the properties of the instrument itself by digital control of both the way it is played and the analysis of the resulting sound. A robot bowing machine activates the string of a cello with a resin-covered glass rod, using a fully varied combination of speeds and pressures. A force recording of the energy transferred to the violin body is then used to train an unsupervised machine learning system, with the goal of discovering what makes the difference between a "good" and "bad" bow stroke - a range of behaviours and resonance regimes that take expert players years to master, described traditionally by terms such as "spiccato", "martelé" and many others.
  
 
Team:
 
Team:

Latest revision as of 08:07, 13 November 2014

The violin family of musical instruments produces its rich effects through an astonishingly complex combination of mechanical and acoustical phenomena, many beyond our current ability to model. It seems likely that the evolution of these instruments, and the cultural and emotional effects they have on listeners, result from this very complexity.

In this project, we are exploring the properties of the instrument itself by digital control of both the way it is played and the analysis of the resulting sound. A robot bowing machine activates the string of a cello with a resin-covered glass rod, using a fully varied combination of speeds and pressures. A force recording of the energy transferred to the violin body is then used to train an unsupervised machine learning system, with the goal of discovering what makes the difference between a "good" and "bad" bow stroke - a range of behaviours and resonance regimes that take expert players years to master, described traditionally by terms such as "spiccato", "martelé" and many others.

Team: