Thesis

Thesis Topic Details

Topic ID:
 1467
Title:
 Modelling Musical Improvisation
Supervisor:
 Mike Bain
Research Area:
 Artificial Intelligence
Associated Staff
Assessor:
 Malcolm Ryan
Topic Details
Status:
 Active
Type:
 Research
Programs:
 CS CE BIOM BINF SE
Group Suitable:
 Yes
Industrial:
 No
Pre-requisites:
 --
Description:
 Modelling musical knowledge to the level of a skilled human is a challenge for
artificial intelligence. Many approaches have been investigated towards this
end, e.g., for modelling composition, stylistic expression and performance
skill. Representations such as grammars and techniques such as neural networks
and genetic algorithms have been used.

A particularly well-studied approach to modelling musical performance has
adopted a rule-based approach. Existing systems can be given a written score
and produce a musically-acceptable performance. Compositional systems that use
sets of rules to produce pieces "in the style of" noted composers have also
been developed.

In this project the goal is to construct a system for jazz improvisation, an
activity which lies somewhere between the tasks of performance and composition.
The hypothesis is that successful improvisation over a set of chord changes can
be achieved by selecting from a given set of note patterns or "lines" and
combining them using suitable "conjunctions" to form a musically-acceptable
performance.

The project will involve a number of stages: research into musical
representation by rules; eliciting a set of lines from a domain expert;
constructing a Markov chain model for linking lines to be played over a given
chord sequence; innovating a novel representation and sequencing technique for
"smoothing" the connection between lines by adding conjunctions.

Applicants must have good programming skills (e.g., in Java).
Fluency in reading musical notation is an advantage.

The project will be jointly supervised by CSE and the School of
Music and Music Education.
Comments:
 Applicants must have good programming skills (e.g., in Java).
Fluency in reading musical notation is an advantage.

The project will be jointly supervised by CSE and the School of
Music and Music Education.
Past Student Reports
  Ryan Martin BRAGANZA in s2, 2009
Modelling Musical Improvisation
  David Jeremiah Joseph GEORGE in s2, 2012
Modelling Musical Improvisation
  Jonathan KAMARAJAN in s2, 2012
Modelling Musical Improvisation
 

Download report from the CSE Thesis Report Library

NOTE: only current CSE students can login to view and select reports to download.