Repository | Book | Chapter
(2015) Mathematics and computation in music, Dordrecht, Springer.
Structural similarity based on time-span sub-trees
Masatoshi Hamanaka, Keiji Hirata, Satoshi Tojo
pp. 187-192
We propose structural similarity of two melodies based on sub-trees from the time-span tree provided by the Generative Theory of Tonal Music. The structural distance of the tree was previously defined and called the "maximal time-span distance," and experimental results showed to some extent a correspondence between the maximal time-span distance and psychological similarity. However, there is a big problem in that almost all pairs of melodies are not similar on the basis of the maximal time-span distance because the definition of the similarity is too strict. Therefore, we attempt to express a melodic structural similarity by using the coincidence rate of time-span sub-trees to weaken the condition for calculating similarity. We have set up three experimental conditions and compared their results.
Publication details
DOI: 10.1007/978-3-319-20603-5_19
Full citation:
Hamanaka, M. , Hirata, K. , Tojo, S. (2015)., Structural similarity based on time-span sub-trees, in T. Collins, D. Meredith & A. Volk (eds.), Mathematics and computation in music, Dordrecht, Springer, pp. 187-192.
This document is unfortunately not available for download at the moment.