TITLE:
Measuring Musical Rhythm Similarity: Further Experiments with the Many-to-Many Minimum-Weight Matching Distance
AUTHORS:
Godfried T. Toussaint, Seung Man Oh
KEYWORDS:
Musical Rhythms, Rhythm Similarity Measures, Many-to-Many Minimum-Weight Matching, Edit Distance, One-to-One Minimum-Weight Matching, Rhythm Perception, Afro-Cuban Rhythms, Middle Eastern Rhythms
JOURNAL NAME:
Journal of Computer and Communications,
Vol.4 No.15,
November
28,
2016
ABSTRACT:
Musical rhythms are represented as sequences of symbols. The sequences may be composed of binary symbols denoting either silent or monophonic sounded pulses, or ternary symbols denoting silent pulses and two types of sounded pulses made up of low-pitched (dum) and high-pitched (tak) sounds. Experiments are described that compare the effectiveness of the many-to-many minimum-weight matching between two sequences to serve as a measure of similarity that correlates well with human judgements of rhythm similarity. This measure is also compared to the often used edit distance and to the one-to-one minimum-weight matching. New results are reported from experiments performed with three widely different datasets of real- world and artificially generated musical rhythms (including Afro-Cuban rhythms), and compared with results previously reported with a dataset of Middle Eastern dum-tak rhythms.