Dissertation Index

Author: Laine, Pauli A.

Title: A Method for Musical Motion Patterns

Institution: University of Helsinki

Begun: July 1994

Completed: April 2000


It is commonly assumed that music consist of separate elements of rhythm, melody and harmony. In examination from generative viewpoint it is possible to see the necessity of the integration of some of the elements. The view is adopted here in which the musical rhythm consist not only of the timing of the events, but also of pitched repeating patterns of elements. It is assumed here that these patterns are result of real or imagined playing movements of the musician. Therefore these patterns are called musical motion patterns in this study. A new method is presented by which it is possible to simulate the musical motion patterns. The method is based on the use of simulated artificial motor neurons connected to form mutually inhibited pairs (MINN). In living systems similarly connected motor neurons form semi-autonomously oscillating systems. These connections are important in wholly or partially non-conscious movements like heartbeat or walking, and contributing also to the aforementioned generation of musical motion patterns of living musicians. The MINN-method is combined with conventional AI-methods so as to make possible the mapping of the generated virtual motion patterns to musical parameters, like pitch or note onset time. Results generated with MINN-method are converted to conventional music notation and compared to movement pattern excerpts found in music of J.S. Bach and L. van Beethoven. The comparison and analysis show that it is possible to generate plausible imitations of musical motion patterns with MINN-method. In the conclusion the applications of the new method in the areas of musicology and computer assisted music composition are discussed.

Keywords: pattern generation, rhythm modelling, algorithmic composition, movement in music


1.1 Related disciplines referred to in this study 6
1.2 Structure of this study 6
- PART 1 - 8
1. Musical motion patterns 8
1.1 Some defining remarks about musical motion patterns 9
1.2 Examples of musical motion patterns 17
1.2 Some music theoretical concepts close to MMP 27
1.3 Features of musical motion patterns 28
2. Music theoretical background 30
2.1 Music analytical theory 30
2.2 Gino Stefani's rhythm theory 31
3 Theories concerning musical movement and motion patterns 34
3.1 Theories about music and rhythm perception 34
3.2 Music imagery, motion and time 35
3.3 Some remarks about improvisation 39
4. Formalized music algorithms 42
4.1 Theoretical background of formalized music algorithms 42
4.2. Formalized modelling methods used in algorithmic music research 43
5. Cognitively oriented algorithms 57
5.1 Studies of musical cognition 57
5.2 Introduction to computational connectionism in music research 58
5.2 Connectionism as a music programming method 62
6. Simulation of pattern generators 67
6.1 Rhythm and pattern generators 67
6.2 Pattern generator simulators 69
- PART 2 -
7. The MINN model 78
7.1 Required features of the model 78
7.2 Main features of the model 79
7.3 Computer implementation 85
7.5 Representing different neural connections using the 'minncl' class 87
7.6 Effect of the parameters 89
7.7 Changing the impulse strength 94
7.8. Visualization and output 94
7.9. Post-processing the time series using music stylistic rules 95
8. Experiments 100
8.1 Interpreting the visual output from the MINN model 100
8.2 Experiments 102
9. Interpreting the results of experiments 137
9.1 Interpreting the results 140
10. Discussion 164
11. Further research and final conclusion 166


Pauli Laine
Mäenrinne 3 E 38
SF-02160 Espoo

     Return to dissertations