Transcript Functional Music

Functional Music
Interim Presentation
Simon McNeilly
Supervisors
Dr. Lloyd Allison
Dr. Jon McCormack
Outline
–
Aims
–
Music analysis
–
Composition of music
–
MIDI
–
Haskore
–
Process
Aims
●
To define some model of musical style
●
To produce a system that can
–
Duplicate the style of some input music
–
Produce new music in that style
–
Compare two styles
–
Determine how well some music fits a style
Musical Analysis
●
Recognition of patterns
–
Repeating elements
–
How they vary
●
Harmony
●
Rhythms
●
Dynamics
Style
●
Part of all music
●
Occurs at multiple levels
●
–
Style of a genre of music
–
Style of a composer
May incorporate any of the information
obtained in the analysis and more
Composition of Music
●
The music composed should have a
purpose
–
Explore possibilities
–
Be enjoyable to listen to
●
●
Use existing music that meets this criteria as a
guide
Use an observed style
Composition of Music
●
A well recognised work in this area is a
program called EMI (Experiments in
Musical Intelligence)
–
Developed by David Cope
–
Maintains a large database of music in a
style
–
Uses a grammar to compose new music
Other Methods of Composition
●
Neural Networks
●
Random composition
–
●
Markov Models
●
Single level
●
Multiple level
Mathematical functions
MIDI
●
●
Widely accepted standard for
representing music
Large databases of freely available
music
MIDI format
●
Tracks
●
Events
●
–
Musical events
–
Meta events
General MIDI
–
Basic MIDI
Haskore
●
Set of modules for representing music in
Haskell
●
Developed at Yale by Paul Hudak
●
Capable of reading and writing MIDI files
Process
●
●
Two parts
–
Analysis of existing music
–
Production of original Music
A model for representing music
–
Will be a simplified model
–
Will have multiple levels
Analysis of Music
●
Pattern recognition problem
–
Recognise melodies
●
–
●
Can be transformed in some way
Recognise structural features
Learn what elements commonly occur in
order to use them in the composition
stage
Composing New Music
●
A matter of loosely copying what is learnt
from input music
–
How it is structured
–
What intervals are common
–
What rhythms are used
Limitations
●
Monophonic only (One sound at a time)
●
Single Instrument
●
Dynamics ignored (Volumes)
●
●
Constant Speed and time signature
assumed
Constant key signature assumed
Conclusion
●
Learn attributes of music like what we
want to produce
–
●
From one or more input songs
Copy and use those attributes
–
Repetition
–
Variation