Thesis Topic Details

Topic ID:
3327
Title:
Variation in Automated Improvisation
Supervisor:
Mike Bain
Research Area:
Artificial Intelligence, Software Engineering, multimedia
Associated Staff
Assessor:
Malcolm Ryan
Topic Details
Status:
Active
Type:
Research
Programs:
SE
Group Suitable:
No
Industrial:
No
Pre-requisites:
--
Description:
Project overview:

Musical performance is a challenging area for scientific analysis. So far,
little is understood about what makes one performance more 'musical' than
another. There has been some initial work on modelling musical style using
rule-based and other approaches. Although this is promising, it is
time-consuming and there is little or no theoretical foundation.
Improvisation, particularly in jazz, has also been the subject of
computational modelling, but this also typically suffers from similar
limitations. Recently, work on the use in an automated improvisation
system of the jazz technique of 'formulaic improvisation' has shown promise.
In this approach, the system makes use of lines, pre-composed musical
fragments, that are stored in a 'repertoire' and combined together at
runtime using a probabilistic model.

However, there are a number of limitations in this approach, one of which
is that although human players can vary the lines used in formulaic
improvisation to adapt them to the piece and the performance, this is not
currently implemented in the system. This project will investigate
approaches to enable musically-motivated variation methods to be used in
the system.

Approach:

Analysis of the music structure of the piece (chart) and potential musical
lines will be necessary. Some way to encode musical 'context' and basic
features of the melodic and harmonic content in a probabilistic rule-based or
grammar-based form. Evaluation will be in terms of how well the system
meets the deign requirements, as well assessment of the outputs in musical
terms.

Skills:

Programming in Java, possibly database design and implementation, simple
user interface design and implentation. Knowledge of
music, both practise and theory.
Comments:
--
Past Student Reports
  Nicholas James GRASEVSKI in s2, 2012
Variation in Automated Improvisation
 

Download report from the CSE Thesis Report Library

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