Date of Award

2002

Degree Type

Thesis

Degree Name

Bachelor of Science (Hons.)

Faculty

Faculty of Computing, Health and Science.

First Advisor

Dr. Philip Hingston

Abstract

The aim of this project is to create software that will give non-musical people the ability to compose convincingly 'real' music in specific musical genres. By 'real' music we mean music which is not obviously "machine generated", recognisable as being of the selected genre, of an aesthetically pleasing quality (a very subjective concept) and usable in a commercial context. To achieve this goal various computational techniques are used including genetic algorithms and finite state automata. The process involves an original, top down approach and a bottom up approach based on previous studies. A novel, skeletal structure is formed to depict a high level description of the song. Sequences of notes are generated based on sample note sequences; taken from songs of the desired style. This study involves generating songs from two different musical genres. Resulting compositions have been objectively assessed by student musicians. To facilitate this the resulting songs were translated into a machine-readable format. Instruments assigned to audition the generated songs were selected from a static, predefined list.

Included in

Composition Commons

Share

 
COinS