臺灣博碩士論文加值系統

Conducting is the act of directing a musical perfor¬mance by way of visible gestures. In a real orchestra, different performers will play their instruments louder, softer, with different artistic expressions, etc. Instrument empha¬sis is the ability to change the aspects of the individual per¬formance of an instrument or section of instruments. With instrument emphasis, a conductor can per¬sonalize a musical piece, so performances sound unique, all due to the majesty of the conductor’s imagination and expres¬siveness. Without such emphasis, performances would be boring and repetitive.
Instrument emphasis has no predefined gestures and can be accomplished in almost any manner. It often involves the gestures used to define tempo or expression, directed towards the desired instrument section. For the human musician, recognizing the emphasis from the conductor becomes trivial with practice. For the computer, the conductor’s gestures are not obvious and computers tend to interpret the hand’s motion simply as its position. To recognize instrument emphasis, the computer needs a means to under¬stand the hand’s motion more like a human would.
In this paper, we present an approach to help the computer understand the hand’s motion through two different but related analyses. We hope to show that by combining the features obtained from the two analyses of the conductor’s hand motion, we can ascertain instrument em¬phasis.
A major achievement of this work is the ability to detect instrument emphasis in real-time. The results of this effort will be combined with other components under construction to build a complete system.

Abstract ii
Acknowledgement iv
List of Tables vii
List of Figures viii
1 Introduction 1
2 Overview of conducting data and conducting trajectory 4
2.1 Conducting data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Conducting trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3 Methodology 7
3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.2 Conducting data de-noise . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2.1 Wavelet shrinkage de-noise . . . . . . . . . . . . . . . . . . . . 9
3.2.2 Real-time wavelet shrinkage de-noise . . . . . . . . . . . . . . 10
3.3 Horizontal analysis of conducting trajectory and feature identication 12
v
3.3.1 Finite states machine model for conducting trajectory abstraction 12
3.3.2 Trajectory change point identification . . . . . . . . . . . . . . 16
3.3.3 Conducting angle identification . . . . . . . . . . . . . . . . . 16
3.4 Vertical analysis of conducting trajectory and feature identification . 18
3.4.1 Vertical conducting trajectory abstraction and analysis . . . . 19
3.4.2 Up beat and conducting zone identification . . . . . . . . . . . 20
3.5 Conducting features discrimination . . . . . . . . . . . . . . . . . . . 22
3.5.1 Real-time outlier detection of the conducting features . . . . . 22
3.5.2 Threshold selection for real-time outlier detection . . . . . . . 23
3.5.3 The locality problem . . . . . . . . . . . . . . . . . . . . . . . 24
4 Evaluation 27
4.1 Datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.3 The input data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5 Conclusions and Future work 34
References 36

[1] Hideyuki Morita and Shuji Hashimoto and Sadamu Ohteru A Computer Music
System that Follows a Human Conductor
[2] Jan Borchers and Eric Lee and Wolfgang Samminger and Max Muhlhauser Per-
sonal orchestra: a real-time audio/video system for interactive conducting, ACM
Multimedia Systems Journal Special Issue on Multimedia Software Engineering.
[3] Eric Lee and Ingo Grull and Henning Kiel and Jan Borchers CONGA: A Frame-
work for Adaptive Conducting Gesture Analysis, NIME 2006 International Conference
on New Interfaces for Musical Expression
[4] Jan Borchers and Aristotelis Hadjakos and Max Mhlhuser MICON: A Music
Stand for Interactive Conducting.
[5] D. L. Donoho and I. M. Johnstone and G. Kerkyacharian and D. Picard Wavelet
Shrinkage: Asymptopia?, J. R. Statist. Soc. B.
[6] Huber, Peter Robust Statistics, Wiley, New York.
[7] Davies, L. and U. Gather The identi
cation of mutiple outliers, J. Amer. Statist. Assoc., 88
[8] Astola, J. and P. Kuosmanen Fundamentals of Nonlinear Digital Filtering, CRC
Press, Boca Raton, New York.