In this paper, we advance an enhanced method for computing Harte et al.’s (2006) Harmonic Change Detection Function (HCDF), which aims to detect harmonic transitions in musical audio signals. Each of the HCDF component blocks is revisited in light of recent advances in harmonic description and transformation. To evaluate our proposal, we compute an exhaustive grid search to compare the multiple proposed algorithms and a large set of parameterizations across four large style-specific musical datasets. Our results show that the newly proposed methods and parameter optimization improve the detection of harmonic changes, by 5.57% (f-score) with respect to previous methods. Furthermore, while guaranteeing recall values at >99%, our other method improves precision by 6.28%.

Authors: Ramoneda, Pedro; Bernardes, Gilberto
Affiliations: University of Zaragoza and University of Porto, Portugal; INESC TEC and University of Porto, Portugal(See document for exact affiliation information.)
AES Convention: 149 (October 2020) Paper Number: 10408
Publication Date: October 22, 2020 Import into BibTeX
Subject: Audio Processing
Permalink: https://www.aes.org/e-lib/browse.cfm?elib=20945

Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!

This paper costs $33 for non-members and is free for AES members and E-Library subscribers.

Learn more about the AES E-Library

E-Library Location:   /conv/149/10408.pdf

Start a discussion about this paper!