Structural Complexity Assessments on Musical Melody and Rhythm
Date Issued
2015
Date
2015
Author(s)
SIMAK, ALEKSANDR
Abstract
There is no definite meaning for music complexity. Current work studied it in terms of complexity of music structure, extending previous research on the complexity of rhythmic structure only. Melody and rhythm this time both encoded within binary tree and the assessment on structural properties of the tree performed. This is the key difference with other studies where complexity assessments performed on music components independently. The concepts of rewriting and L-Systems used to extract repeatable patterns from the tree and form tree generating context-sensitive grammar. Elementary notes transition represented as grammar’s rewriting rules. Rewriting rules classification transforms context-sensitive grammars to context-free stochastic variant. More complex tree has more complex generating grammar, the final assessment score is well-known complexity (or entropy) of context-free grammar. Current work updated the method with new binary tree. Redefined tree is capable to store arbitrary content inside the nodes, thus encoding dissimilar features such as melody and rhythm within its structure and nodes simultaneously. Notions of similarity for rewriting rules were updated accordingly. Better algorithm for rewriting rules classification was proposed and approbated during the study. Finally, minor theoretical conform issues of notations were accurately justified. Updated method was approbated with two datasets: drums exercises and Barlow dictionary of 10,000 classical themes. Empirical results revealed enough method sensitivity to detect atypical samples within the corpus, discriminate samples by their relative complexity and score the comprehensiveness of particular musical or personal style.
Subjects
music complexity
structural complexity
barlow dictionary
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-104-D98922040-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):e7355c3ce0b681eecd8d7e4ad59e64f7