Title: Music image processing
1Music image processing
- Tim Bell
- Department of Computer Science and Software
Engineering - University of Canterbury, Christchurch, New
Zealand
2With
- David Bainbridge (Waikato)
- Richard Lobb
- Dave Blizzard (Portland, OR)
- Richard Green
- John McPherson
- Karen Lin
- Annabel Church
- Simon Glass
3Overview
- OMR
- Digital music stand
- Page turning and image size
- Fast capture of music
- Music classification
4The vision...
- All music available on the web
- as score, recording and MIDI
- search by name, composer, lyrics, phrase,
similarity, genre, novelty - remunerate those responsible
- culturally sensitive access
5Barlow and Morgenstern 1949
6Barlow and Morgenstern 1949
7Barlow and Morgenstern 1949
8Activities with music
- Composing
- Arranging
- Performance
- Teaching
- Musicology
- Recording
- Accompanying
- Transcribing...
9Digital music problems
- Cost of hardware and software
- Viewing music on small screen
- Loss of inspiration and creativity
- Loss of efficiency
- Learning curve
- Software compatibility
10Problems with paper
- Pen or pencil?
- Indexed retrieval
11Visual display
Digital image
Human memory
Digital semantic
Audio
12Visual display
Digital image
memorised oral tradition original ideas
Human memory
Digital semantic
Audio
13Visual display
Digital image
Human memory
live performance mp3, wav, cd video?
Digital semantic
Audio
14Visual display
Digital image
sheet music screen
Human memory
Digital semantic
Audio
15Visual display
Digital image
sheet music screen
Human memory
Digital semantic
Audio
16Visual display
Digital image
sheet music screen
Human memory
Digital semantic
Audio
17Visual display
Digital image
sheet music screen
Human memory
Digital semantic
Audio
18Visual display
Digital image
sheet music screen
Human memory
Digital semantic
Audio
19Visual display
Digital image
D (lick 1) Dmin5 Look at me
now, will I ever learn? D (lick
2) Dmin5 G I dont know how but I
suddenly lose control.
sheet music screen
Human memory
Digital semantic
Audio
20Visual display
Digital image
Live performance
Human memory
Digital semantic
Audio
21Visual display
Digital image
read
Live performance
Human memory
Digital semantic
Audio
22Visual display
Digital image
Live performance
Human memory
play (interpret)
Digital semantic
Audio
23Visual display
Digital image
Transcription
Human memory
Digital semantic
Audio
24Visual display
Digital image
Transcription
Human memory
listen
Digital semantic
Audio
25Visual display
Digital image
write
Transcription
Human memory
Digital semantic
Audio
26Visual display
Digital image
BMP, GIF, JPEG
Human memory
Digital semantic
Audio
27Visual display
Digital image
scanner, camera
Human memory
Digital semantic
Audio
28Visual display
Digital image
print, display
Human memory
Digital semantic
Audio
29Visual display
Digital image
Human memory
MIDI NIFF MUSICXML GUIDO
Digital semantic
Audio
30Visual display
Digital image
Render (Sibelius, Lime, Guido, Tex etc.)
Human memory
Digital semantic
Audio
31Visual display
Digital image
OMR
Human memory
Digital semantic
Audio
32Visual display
Digital image
Human memory
Synthesis (audio rendering)
Digital semantic
Audio
33Visual display
Digital image
Human memory
Audio analysis (monophonic, polyphonic)
Digital semantic
Audio
34Visual display
Data entry
Digital image
Human memory
Digital semantic
Audio
35Visual display
Digital image
Weak links
Human memory
Digital semantic
Audio
36Visual display
Digital image
Labour intensive links
Human memory
Digital semantic
Audio
37Visual display
Digital image
Human memory
Digital semantic
Audio
38Visual display
Digital image
QBH
Human memory
Digital semantic
Audio
39Operations on music
40Visual display
Digital image
Compose Arrange/orchestrate Rehearse React
Human memory
Digital semantic
Audio
41Visual display
Digital image
Library (personal, shared) Music stand (rehearse,
perform)
Human memory
Digital semantic
Audio
42Visual display
Digital image
Intermediate form Archive
Human memory
Digital semantic
Audio
43Visual display
Suitable for transposition part
splitting reduction searching theme
detection accompaniment performance following
Digital image
Human memory
Digital semantic
Audio
44Visual display
Digital image
listening recording studio analysis thumbnail back
ing
Human memory
Digital semantic
Audio
45Optical Music Recognition
46(No Transcript)
47(No Transcript)
48Wabot-2
- 1980-1984
- Read simple score
- Heavy processing requirements
49(No Transcript)
50Staff line removal/identification
- Horizontal projection
- Vertical slices
- Wobble/track
- Chords
- Template
51(No Transcript)
52Horizontal projection
53Piece at an angle
54Vertical slices
55Vertical slices
56Wobble/track
57Object location
- Fragmentation
- Superimposed
- Touching objects
58Identifying objects
- Flood fill
- Template matching
- Hough transform
59Constructing musical features
- Grammars
- Decision tree
- Rules
60Musical semantics
- Treble clef determines pitches
- Accidentals change pitch
- Time signature changes note lengths
61Commercial systems
- Sharpeye
- Vivaldi
- Neuratron PhotoScore
- and more
62Is 96 recognition good enough?
- One mistake in 24 notes
- No interpretation if playing music
- Time to set up, train and correct greater than
typing?
63Not just notes
64(No Transcript)
65Visual display
Digital image
Optical music recognition in practice!
Human memory
Digital semantic
Audio
66The gulf of interpretation
- Classical tempo and dynamics
- Jazz improvisation
- Rock style (e.g. syncopation, articulation)
- Figured bass
- Cadenzas
- MIDI vs. Orchestra
67Interpretation
68Interpretation
69Visual display
Digital image
Pen-based music data entry
Human memory
Digital semantic
Audio
70(No Transcript)
71(No Transcript)
72Coloured staveline removal
- Scanned as RGB
- Convert to HSV and CMYK
- V indicates colour
- K indicates pencil, black pen
73(No Transcript)
74(No Transcript)
75Mis-classified images
76Discussion?