Interview with Harrald
Quote:
Your role as computer musician is the most visually inconspicuous of the trio, yet you have just as much control over the direction the performance will take. How do you exercise that control? How much of the decision making is left to your chosen computer program, and how does it lead the performance as an independent musician without your input? What informs your choice of program, and how do you tailor it to suit the needs of your performances?
One of the unique aspects of our performance is that through using Artificial Intelligence the computer is an active collaborator in shaping the performance, and while I can influence its musical output though the samples selected for a given performance, beyond that the computer is reasonably autonomous. I mostly respond to its output, just as I would to the musical directions of Derek and Chris, mixing the 8 channels of audio it generates as the performance unfolds. I can influence the computer by telling it I like or don’t like what it is doing, but this does not directly trigger specific samples, just influences how volatile the computer’s musical selections will be.
The ENSEMBLE system – an interactive computer music environment I created, and have been performing with and tweaking since 2007 – aims to model improvisatory behavior rather than musical structures or processes using Game Theory. The system is an ensemble of virtual software agents that ‘make decisions’ based on strategies that allow the system to mimic some of the conscious and unconscious strategies of human improvisers. The virtual ensemble can work autonomously to create algorithmic compositions, or interactively in live performance.
In interactive mode, the system is essentially a musical game. Live performers interact with the system (either through a pitch tracking system in the case of Derek, or via the mouse for me) and compete with the virtual ensemble for control of the performance. The competitive nature of the system is important as rather than just following the improvisations of the live performers, the virtual ensemble is responsive, either reinforcing the live performers initiatives or subverting them depending on how the performers have interacted with the system previously. If you fair poorly in the game, the system becomes quite unresponsive to your input, and does it’s own thing!
The system has been used in a wide range of settings from generating film soundtracks, to installations, audience mediated performance via wireless and even an orchestral work. The system is modular, meaning that you can expand its capabilities to cover different situations easily through adding new software components to the existing system. I program in the MaxMSP programming environment.
As the behavior of the agents mimics real-life behaviors (often the agents flick between channels like small children fighting over which TV channel they are going to watch!) it is interesting working with the system in that it gives the you the impression of a living entity, hence I often perform under the title of ‘AI Hander’, along the lines of the Elephant Handler, or Lion Tamer in the Circus.
|
I just quite liked the Q and A here, particularly interesting for me at least.
The entire interview is at
De La Catessen, as well as interviews and ideas presented about the other live acts.
Just CTRL+F 'MiniMax' and the interview should be shortly thereafter.
I don't know how long it will be up for, but enjoy nonetheless.