IN THE CONTEXT OF ELECTRONIC PUBLISHING
BY
NICHOLAS PAUL CARTER, G.R.S.M. (Hons.), L.R.A.M.
A Thesis submitted to the Faculties of Science and Human Studies at the University of Surrey in partial fulfilment of the requirements for the degree of Doctor of Philosophy.
Depts. of Physics and Music
University of Surrey
February 1989
Abstract
Computers are used to manipulate music in various forms, for example digital sound recordings, digitized images of printed scores and music representational language (M.R.L.) encodings. This work is concerned with producing M.R.L. data automatically from existing printed music scores.
A review of work undertaken in the field of manipulating printed music by computer is provided. This shows that software which permits production of high-quality scores is commercially available, but the necessary data has to be entered using some form of keyboard, possibly in conjunction with a pointing device. It is desirable, for reasons detailed in this work, to be able to convert the musical information contained in the enormous quantity of existing music into computer-readable form. The only practical method for achieving this is via an automatic system. Such an automatic system must cope with the variations in format, content and print-quality of existing scores.
Background material relating to previous work on pattern recognition of various types of binary image is included, with a section covering the subject of automatic recognition of printed music. An original system for automatic recognition of printed music developed by the author is described. This is designed to be widely applicable and hence is, in effect, omnifont and size-independent, with significant tolerance of noise, limited rotation, broken print and distortion. Numerous illustrations
ii
showing the application of the system are included, together with proposals for future areas of development.
iii
Contents
Acknowledgements vii
Chapter 1 INTRODUCTION 1
Chapter 2 THE ACQUISITION, REPRESENTATION AND
RECONSTRUCTION OF PRINTED MUSIC BY COMPUTER: A REVIEW 5
2.1 INTRODUCTION 5
2.2 ACQUISITION 6
2.2.1 Automatic pattern recognition of printed music 7
2.2.2 Operator input 8
2.2.2.1 Directly-connected instrument 8
2.2.2.2 Soundtrack analysis 9
2.2.2.3 The compromise solution to the input problem 10
2.2.2.4 Practical assessment of input methods 11
2.3 REPRESENTATION 15
2.3.1 Existing M.R.L.'s 18
2.3.1.1 DARMS: Digital Alternate Representation
of Musical Scores 19
2.3.1.2 MUSTRAN: Music Translater 20
2.3.1.3 ALMA: Alphanumeric Language
for Music Analysis 21
2.3.2 M.R.L. structure 22
2.3.3 The Editor 22
2.3.4 Problems concerning score representation
in the context of printing 24
2.4 RECONSTRUCTION
iv
2.5 APPENDIX (Table of music printing systems) 31
Chapter 3 PATTERN RECOGNITION OF BINARY IMAGES 50
3.1 BACKGROUND 50
3.2 PATTERN RECOGNITION OF BINARY IMAGES 51
3.3 PREVIOUS WORK ON PATTERN RECOGNITION
OF PRINTED MUSIC 61
3.3.1 Background 62
3.3.2 Aims 63
3.3.3 Acquisition 64
3.3.4 Staveline finding and segmentation 68
3.3.5 Grammar and syntax 73
3.3.6 Object (symbol) formation and recognition 75
3.3.7 Interaction and output 83
3.3.8 Error-checking/correction and future aims 86
Chapter 4 A SYSTEM FOR THE AUTOMATIC RECOGNITION OF
PRINTED MUSIC : TECHNIQUES 89
4.1 ACQUISITION OF THE BINARY IMAGE 89
4.2 EXPERIMENTAL SEGMENTATION 90
4.3 SEGMENTATION 93
4.4 NOISE REMOVAL 98
4.5 FILAMENTS AND STAVELINE FINDING 99
4.6 FILAMENT STRINGS AND STAVE FINDING 101
4.7 SIMPLIFICATION OF THE TRANSFORMED LAG 103
4.8 OBJECT FORMATION 107
4.9 SYMBOL ORDERING 107
4.10 RECOGNITION 108
4.11 ANALYSIS OF BEAMED NOTE GROUPS 114
v
4.12 SYSTEM PARAMETERS 122
Chapter 5 A SYSTEM FOR THE AUTOMATIC RECOGNITION OF PRINTED MUSIC : WORKED EXAMPLES 124
5.1 INTRODUCTION 124
5.2 WORKED EXAMPLES 125
Chapter 6 CONCLUSIONS 150
Postscript 156
Bibliography 157
vi
Acknowledgements
Grateful thanks are extended to my supervisors, Dr. Dick Bacon and Dr. Tom Messenger, and also to Mr. Andrew Potter (Head of Music, Oxford University Press) and Mr. David Munro, for their continual support and encouragement.
Thanks are also due to the following :-
Racal Information Technology Developments Ltd., for the use of their modified Canon 9030 laser copier, which now forms the basis of their REOS document storage system.
Oxford University Press, the Performing Right Society and the Leverhulme Trust, whose sponsorship made this research possible.
Peters Edition Ltd., Schott & Co. Ltd., Fentone Music Ltd. (agents for Friedrich Hofmeister Musikverlag) and Boosey & Hawkes Music Publishers Ltd., who kindly gave permission for reproduction and manipulation of their copyright material.
I would also like to thank my colleagues in the Physics and Music Departments for their contributions throughout the duration of this research.
This work is dedicated to my father and mother.
vii