Information Technology Reference
In-Depth Information
mechanical process that ultimately delivers an outgoing audio stream, rendered by
a digital-to-analogue audio convertor.
The objective of a mechanical thinking brain is very far from realisation. The
pending issue, therefore, is how a machine could emulate human performance con-
vincingly enough that companion improvisers, and listeners, would accept the Live
Algorithm as a contributing and creative group member with the same musical sta-
tus as any other performer. The overarching aim is to extend the possibilities of
music performance, achievable by the challenges of interacting creatively with a
mechanical process, and by the exploitation of the pristine world of algorithmic
patterning.
To achieve this, Live Algorithms should be able act responsively, proactively and
appropriately without direct intervention, and contemporary methods in Artificial
Intelligence, applied in real-time, offer suitable opportunities for rising to this chal-
lenge. The resulting systems, those that truly achieve this goal or at least achieve
steps towards it, differ substantially to the norms of computer-as-instrument (with a
fundamental reliance on human agency) and computer-as-score, in which a design-
ers intentions are encoded as a set of rules or instructions, comparable to a musical
score.
This chapter will examine several aspects of Live Algorithms and improvised
music. In order to set the scene, a description of improvised music is given, speci-
fying four attributes that we ascribe to human performers, and by implication, to a
Live Algorithm: autonomy, novelty, participation and leadership (Sect.
6.2
).
A formal specification is given that helps situate Live Algorithms in a wider tax-
onomy of computer music systems, based on a modular decomposition into analysis,
patterning/reasoning and synthesis components (Sect.
6.3
). A Live Algorithm is de-
fined as a system in which these three elements are present, interconnected, absent
of a human controller, and such that the above four attributes are satisfied. Several
possible configurations of these modular elements (with or without a human con-
troller) are considered, relating to a number of existing computer music practices.
The core of the Live Algorithm, the patterning element,
f
, is considered in greater
detail and a dynamical systems approach to their design is outlined.
The chapter continues with a discussion concerning the experience of perform-
ing with a Live Algorithm (or of hearing a Live Algorithm performance) in terms
of the nature of performance interaction, and the possibilities of human-machine
interaction in the near future (Sect.
6.4
). Examples of simple behaviours that might
be expected to occur are provided. Human performers operate in an extended con-
text that lies beyond any single performance; they are subject to social and cultural
forces that shape and inform their own approach to music. These social aspects and
consequent implications for Live Algorithms are discussed.
Cumulatively, these ideas are offered as theoretical tools for the field of Live
Algorithm research. Having presented them in very abstract terms, we turn to dis-
cussing a number of prototypes that we believe conform to the structure of a Live
Algorithm (Sect.
6.5
).
The chapter closes with an account of some possible directions for Live Algo-
rithm research.
