Metrical strength and contradiction in Turkish makam music André Holzapfel, Barı¸s Bozkurt Bahçe¸sehir University, Istanbul, Turkey
13 July, 2012
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
1 / 16
1
Introduction
2
Setup
3
Results
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
2 / 16
Outline
1
Introduction
2
Setup
3
Results
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
3 / 16
Western music: Metrical strength and syncopation Metrical strength Various studies exist that evaluate how note position and duration support the underlying meter (e.g. Palmer & Krumhansl 1990).
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
4 / 16
Western music: Metrical strength and syncopation
Syncopation New Harvard Dictionary of Music Syncopation is a temporary contradiction of the prevailing meter. Either a different meter is superimposed (hemiola), or single note events are placed in unusual manner. Only few computational methods were proposed, concentrating on event syncopation.
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
5 / 16
Western music: Metrical strength and syncopation
Huron and Ommen (2006) Based on manual annotation, popular American music was examinated for syncopation.
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
6 / 16
Western music: Metrical strength and syncopation
Sioros et al.(2012) A set of popular polyphonic music (MIDI) was examined for the distribution of syncopation.
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
7 / 16
Meter in Turkish makam music Symbolic description of the usul Aksak
9 8 DUM TE KE DUM TEK TEK
Percussion example, Composition example Usul have lengths from 2 up to 48 or more (example Aksak : length 9) Separation into short (2-15) and long usul. The denominator (mertebe) specifies the tempo class of the composition. Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
8 / 16
Outline
1
Introduction
2
Setup
3
Results
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
9 / 16
Experimental setup
Dataset CLASS AKSAK CURCUNA DÜYEK SEMAI SOFYAN ˘ TÜRK AKSAGI
NSongs 64 57 47 22 60 38
NNotes 28440 23363 17093 8376 32139 12972
Beats 9 10 8 3 4 5
Mertebe 8 8 8 4 4 8
All songs contain the main melody in MIDI format.
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
10 / 16
Experimental setup
Experimental setup The melodies of all compositions are sampled at a resolution of 1/16-notes. For each usul, the number of note events and their average durations are determined for each bin at this resolution. Using the NLHL-p algorithm proposed by Sioros et al. (2012), “syncopated” events are detected. Then, their distribution among the 1/16-note bins is determined.
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
11 / 16
Outline
1
Introduction
2
Setup
3
Results
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
12 / 16
Note Location
3
3
2
2
Weight
Weight
Correlation coefficients 1 0
2
4
6
8 10 12 Weight location (1/16)
14
16
1 0
18
3
3
2.5
2.5
2 1.5 1 0.5 0
2
4
6
8
10 Location (1/16)
12
14
16
(c) Aksak: Locations
1
2
3
4
5 6 7 8 Weight location (1/16)
9
10
11
12
(b) Semai: Theory Normalized Count
Normalized Count
(a) Aksak: Theory
18
2 1.5 1 0.5 0
1
2
3
4
5
6 7 Location (1/16)
8
9
10
11
12
(d) Semai: Locations
CLASS AKSAK CURCUNA DÜYEK SEMAI SOFYAN ˘ TÜRK AKSAGI
ro 0.859 0.826 0.762 0.943 0.572 0.819
Correlation coefficients tend to be lower than the ones reported for Western music. Weaker beats obtain more weight than would be expected from theory. Usuls give a sparse description of the theoretical weights. Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
13 / 16
Note durations
3
3
2
2
Weight
Weight
Correlation coefficients 1 0
2
4
6
8 10 12 Weight location (1/16)
14
16
1 0
18
1
(e) Aksak: Theory
4
5 6 7 8 Weight location (1/16)
9
10
11
12
6 Mean Duration (1/16)
Mean Duration (1/16)
3
(f) Semai: Theory
4 3 2 1 0
2
2
4
6
8
10 Location (1/16)
12
14
16
(g) Aksak: Durations
18
4
2
0
1
2
3
4
5
6 7 Location (1/16)
8
9
10
11
12
(h) Semai: Duration
CLASS AKSAK CURCUNA DÜYEK SEMAI SOFYAN ˘ TÜRK AKSAGI
rD 0.903 0.883 0.792 0.969 0.818 0.749
Correlation coefficients from durations are higher than the ones for note locations.
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
14 / 16
Metrical contradiction Syncopation localizations and lengths for usul Aksak and Düyek 800 Frequency
Frequency
300 200 100 0
2
4
6
8 10 12 Center location (1/16)
14
16
200 2
4
6
8 10 Center location (1/16)
12
14
16
2
4
6
8 10 Syncopation length (1/16)
12
14
16
600 Frequency
Frequency
400
0
18
300 200 100 0
600
2
4
6
8 10 12 Syncopation length (1/16)
14
16
18
400 200 0
(i) Aksak
(j) Düyek
Findings: Some usul have systematic syncopation (Düyek ). In contrary to Western music, this syncopation tends to be concentrated in the first half of the meter. Syncopated events tend to have a larger duration than for Western music. Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
15 / 16
Conclusion
Note location and duration support the meter in makam music. The correlation at least for note locations is lower as correlation reported for Western music. Even though the term syncopation hardly ever encountered in Makam music literature, it seems to play an important role in supporting the meter as well. An usul classification system based on learned onset patterns is likely to give good results.
Holzapfel et al. (BAU)
Metrical strength and contradiction
13 July, 2012
16 / 16