Dear all,
I would be most grateful if someone of you could proof-read the
Oz-related parts of my PhD thesis. The thesis abstract can be found at
the end of this mail.
Please contact me off-list.
Thank you very much indeed!
Best,
Torsten
Abstract Draft:
Constraint programming gained much interest in computer aided
composition, because it allows to express explicit compositional
knowledge in a declarative way. In this field, the user states a music
theory and the computer generates music which complies this theory.
Music constraint programming is style-independent and is well-suited
for highly complex theories (e.g. a fully-fledged theory of harmony). A
theory is formulated as a constraint satisfaction problem (CSP) by a
set of rules (constraints) applied to a music representation in which
some aspects are expressed by variables (unknowns). A generic music
constraint system predefines concepts shared by many theories (e.g.
duration, pitch, note, chord) and that way greatly simplifies the
definition of musical CSPs.
This research presents the design, usage, and evaluation of a generic
music constraint system called Strasheela. Compared with existing music
constraint systems, Strasheela is more generic, that is Strasheela
supports a considerably larger set of musical CSPs or makes their
definition more concise -- while still performing reasonably efficient.
Strasheela is more generic, because it is more programmable. In
particular, three fundamental components are more programmable in
Strasheela: the music representation, the rule application mechanism,
and the search process.
Strasheela features an highly expressive symbolic music representation.
The representation stores diverse information explicitly by a set of
score objects, their attributes, and the hierarchic nesting of these
objects (e.g. objects such as a note and a voice, with attributes such
as the pitch of a note and the duration of a voice, where a sequence of
notes is contained in a voice). Any explicit information available in
the score is accessible from any score object and can be used to obtain
derived information (e.g. the interval between the pitches of two
notes). Explicit and derived score information can be expressed by
variables. Such information is unknown in the CSP definition and can be
constrained. A rich data abstraction interface allows convenient access
to explicitly stored or derived information. The basic interface (e.g.
type-checkers or attribute accessors) is complemented by generic
higher-order abstractions (e.g. collecting all objects in the score
hierarchy which fulfill a user-defined condition), and abstractions
which express specific musical knowledge (e.g. accessing neighbouring
notes in a voice, or notes which are simultaneous in the score
hierarchy). The user can incrementally extend the music representation
(e.g. defining new score types by class inheritance).
This research proposes a rule formalism which combines convenience and
full user control to express which sets of variables in the music
representation are constrained by a given rule. A Strasheela rule is a
first-class abstraction and a rule application mechanism is a
higher-order abstraction implementing an arbitrary traversal across the
score to apply a given rule to the score. This text presents rule
application mechanism examples which are suitable for a large set of
musical CSPs and reproduces the application mechanisms of important
existing systems.
Strasheela is founded on a constraint programming model based on the
notion of computation spaces. This model makes the search process
programmable at a high-level. Strasheela customises the space-based
constraint model for music constraint programming by preserving its
full programmability. The Strasheela user can optimise the search for a
particular constraint satisfaction problem by programming the decision
making process conducted during search (the branching heuristic, the
distribution strategy) -- independent of the problem definition. This
decision making process can conveniently exploit any information
available in the score at the time of the decision. Special
distribution strategies for efficiently solving musical CSPs including
complex polyphonic problems are proposed.
This research provides a prototype implementation which comes with
examples and documentation.
--
Torsten Anders
Sonic Arts Research Centre • Queen's University Belfast
Frankstr. 49 • D-50996 Köln
Tel: +49-221-3980750
www.torsten-anders.de
_________________________________________________________________________________
mozart-users mailing list
[email protected]
http://www.mozart-oz.org/mailman/listinfo/mozart-users
