Aligned hierarchies for sequential data

Kinnaird, Katherine M.

doi:10.1349/ddlp.1256

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These aligned hierarchies when applied to music-based data sequences can be used for visualization or for similarity tasks within MIR like the fingerprint task (that seeks to find all copies of a query song that are the same recording of the same piece performed by the same musical group) (Kinnaird, 2016). The aligned hierarchies can also be post-processed to address additional tasks such as the cover song task (that seeks to find different recordings of the same piece of music either by the original artist or another one) (Kinnaird, 2018;McGuirl et al, 2018;Savard et al, 2020). A drawback of Kinnaird's code, however, is that the original code was written in MATLAB, which can only be used with a license and hence is not broadly accessible.…”

Section: Statement Of Needmentioning

confidence: 99%

repytah: An Open-Source Python Package for Building Aligned Hierarchies for Sequential Data

Jia,

Carpenter,

Tran

et al. 2023

JOSS

View full text Add to dashboard Cite

We introduce repytah, a Python package that constructs the aligned hierarchies representation that contains all possible structure-based hierarchical decompositions for a finite length piece of sequential data aligned on a common time axis. In particular, this representation-introduced by Kinnaird (2016) with music-based data (like musical recordings or scores) as the primary motivation-is intended for sequential data where repetitions have particular meaning (such as a verse, chorus, motif, or theme). Although the original motivation for the aligned hierarchies representation was finding structure for music-based data streams, there is nothing inherent in the construction of these representations that limits repytah to only being used on sequential data that is music-based.The repytah package builds these aligned hierarchies by first extracting repeated structures (of all meaningful lengths) from the self-dissimilarity matrix (SDM) for a piece of sequential data. Intentionally repytah uses the SDM as the starting point for constructing the aligned hierarchies, as an SDM cannot be reversed-engineered back to the original signal and allows for researchers to collaborate with signals that are protected either by copyright or under privacy considerations. This package is a Python translation of the original MATLAB code by Kinnaird (2014) with additional documentation, and the code has been updated to leverage efficiencies in Python.

show abstract

Section: Statement Of Needmentioning

confidence: 99%