1. Algorithmic composition in Haskell
s1180004 Hajime Ito
University of Aizu
Language Processing Systems Laboratory
Advisor: Taro Suzuki
2013/12/18

2. Background of Computer music
Today, the automatic composition software is very
common.
Anyone can obtain it and compose original music free
of charge.
This is a benefit of the studies of the music composition
algorithm.
Ｔhere are infinity variations of the human made music.
If we want to realize such music in algorithmic
composition, further research should be needed in the
future.

3. The research being investigated
 Six main point of the composition
・Music Adam Alpern, Techniques for Algorithmic Composition of Hampshire College, 1995
 Composition algorithm in other languages (mainly C and Java)
・Mary Simoni, Algorithmic composition: a gentle introduction to music composition
using common LISP and common music. The University of Michigan.
・Algorithmic Composition Methods for Breakbeat Science, Nick Collins, Centre for
Electronic Arts
 Music information processing in Haskell
Jeremy Gibbons, fun of functional programming
 I can not find any Haskell’s research for music composition.

4. Existing problems in Computer music
 Lack of the usability
Once a music is generated, user cannot change phrase partially.
User cannot select music instruments for each part.
 Weak point of the software come from the language programmer used
Memory reservation
Haw to treat infinity lists
 Require technical knowledge both music and programming field.

5. Significance of this research
Programming language Haskell approach
○Haskell is suitable for creating and operation of the infinity list.
・In fact, most of the existing algorithm treat music data as
infinity list.
 Haskell can handle music algebraically.
○Key change operation will realize only a few source code.
Free order for generating phrase thanks to Lazy evaluation.
○It resembles the composition which human done.

6. User
Researching area
Chord
Instrument
Rhythm
etc.
Automatic Composition software
Output as MIDI file
♪
MIDI
Generate music
Software
synthesizer
Play music

7. Strong points of Haskell algorithm
 In Phrase generating Algorithm
○ Generate phrase from the place that user want to need.
 In Repetition Management Algorithm
○ Due to lazy evaluation, Haskell can generate and provide
repeated phrases as much as needed.
 In Modifying change Algorithm
○ Easy to realize such operations: Key change, Volume control
・Kindly and useful for users and developers.

8. Research report 1
In phase of main phrase generation
Select chord progressions from famous patterns.
Require dedicated algorithm for each beats
(e.g. 1/3, 1/4, 1/5)
In phase of drum part generation
Drum part is less necessary for the randomness than
other instrument parts.
・Simple repetition of specific phrase is roughly allowed.

9. Research report 2
In phase of sub phrase generation
Put some notes which belong chord of main part.
It result in noiseless melody.
In phase of sub-sub phrase generation
I cannot realize this algorithm.

10. Research report 3
 How to generate natural phrase?
Rules of the limit of the phrase
Length of one phrase is free, but every 4 section must have 2 ~ 8 phrases.
Rules of the pitches in every phrase (about common music)
If program select a pitch with greatly spaced from the last one,
the next pitch should be selected within less spaced ones.

11. Future extensibility
I have to improve...
Variations of generated music
Algorithm which not restricted by types of beats.
Algorithm for each instrument.
Algorithm realize various performance techniques.
Useful GUI for general public.
I want to involve the people who were not interested in
computer music until now. Then, I intend to solve these tasks.

12. Conclusion
 The algorithmic composition in Haskell was surely significant.
Convenience for programmers
・memory management
・concise description
・(easy to setup environment)
We have to improve both algorithm and software as
one system as well as general user can use it.