Cover song identification with metric learning using distance as a feature

Hoon Heo, Hyun Woo Kim, Wan Soo Kim, Kyogu Lee

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Most of cover song identification algorithms are based on the pairwise (dis)similarity between two songs which are represented by harmonic features such as chroma, and therefore the choice of a distance measure and a feature has a significant impact on performance. Furthermore, since the similarity measure is query-dependent, it cannot represent an absolute distance measure. In this paper, we present a novel approach to tackle the cover song identification problem from a new perspective. We first construct a set of core songs, and represent each song in a high-dimensional space where each dimension indicates the pairwise distance between the given song and the other in the pre-defined core set. There are several advantages to this. First, using a number of reference songs in the core set, we make the most of relative distances to many other songs. Second, as all songs are transformed into the same high-dimensional space, kernel methods and metric learning are exploited for distance computation. Third, our approach does not depend on the computation method for the pairwise distance, and thus can use any existing algorithms. Experimental results confirm that the proposed approach achieved a large performance gain compared to the state-of-the-art methods.

Original languageEnglish
Title of host publicationProceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017
EditorsZhiyao Duan, Douglas Turnbull, Xiao Hu, Sally Jo Cunningham
PublisherInternational Society for Music Information Retrieval
Pages628-634
Number of pages7
ISBN (Electronic)9789811151798
Publication statusPublished - 2017 Jan 1
Externally publishedYes
Event18th International Society for Music Information Retrieval Conference, ISMIR 2017 - Suzhou, China
Duration: 2017 Oct 232017 Oct 27

Publication series

NameProceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017

Conference

Conference18th International Society for Music Information Retrieval Conference, ISMIR 2017
CountryChina
CitySuzhou
Period17/10/2317/10/27

Fingerprint

Distance education
Distance Learning
Song

ASJC Scopus subject areas

  • Music
  • Information Systems

Cite this

Heo, H., Kim, H. W., Kim, W. S., & Lee, K. (2017). Cover song identification with metric learning using distance as a feature. In Z. Duan, D. Turnbull, X. Hu, & S. J. Cunningham (Eds.), Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017 (pp. 628-634). (Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017). International Society for Music Information Retrieval.

Cover song identification with metric learning using distance as a feature. / Heo, Hoon; Kim, Hyun Woo; Kim, Wan Soo; Lee, Kyogu.

Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017. ed. / Zhiyao Duan; Douglas Turnbull; Xiao Hu; Sally Jo Cunningham. International Society for Music Information Retrieval, 2017. p. 628-634 (Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Heo, H, Kim, HW, Kim, WS & Lee, K 2017, Cover song identification with metric learning using distance as a feature. in Z Duan, D Turnbull, X Hu & SJ Cunningham (eds), Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017. Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017, International Society for Music Information Retrieval, pp. 628-634, 18th International Society for Music Information Retrieval Conference, ISMIR 2017, Suzhou, China, 17/10/23.
Heo H, Kim HW, Kim WS, Lee K. Cover song identification with metric learning using distance as a feature. In Duan Z, Turnbull D, Hu X, Cunningham SJ, editors, Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017. International Society for Music Information Retrieval. 2017. p. 628-634. (Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017).
Heo, Hoon ; Kim, Hyun Woo ; Kim, Wan Soo ; Lee, Kyogu. / Cover song identification with metric learning using distance as a feature. Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017. editor / Zhiyao Duan ; Douglas Turnbull ; Xiao Hu ; Sally Jo Cunningham. International Society for Music Information Retrieval, 2017. pp. 628-634 (Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017).
@inproceedings{239af11e7b954cda9889e1bb53259786,
title = "Cover song identification with metric learning using distance as a feature",
abstract = "Most of cover song identification algorithms are based on the pairwise (dis)similarity between two songs which are represented by harmonic features such as chroma, and therefore the choice of a distance measure and a feature has a significant impact on performance. Furthermore, since the similarity measure is query-dependent, it cannot represent an absolute distance measure. In this paper, we present a novel approach to tackle the cover song identification problem from a new perspective. We first construct a set of core songs, and represent each song in a high-dimensional space where each dimension indicates the pairwise distance between the given song and the other in the pre-defined core set. There are several advantages to this. First, using a number of reference songs in the core set, we make the most of relative distances to many other songs. Second, as all songs are transformed into the same high-dimensional space, kernel methods and metric learning are exploited for distance computation. Third, our approach does not depend on the computation method for the pairwise distance, and thus can use any existing algorithms. Experimental results confirm that the proposed approach achieved a large performance gain compared to the state-of-the-art methods.",
author = "Hoon Heo and Kim, {Hyun Woo} and Kim, {Wan Soo} and Kyogu Lee",
year = "2017",
month = "1",
day = "1",
language = "English",
series = "Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017",
publisher = "International Society for Music Information Retrieval",
pages = "628--634",
editor = "Zhiyao Duan and Douglas Turnbull and Xiao Hu and Cunningham, {Sally Jo}",
booktitle = "Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017",

}

TY - GEN

T1 - Cover song identification with metric learning using distance as a feature

AU - Heo, Hoon

AU - Kim, Hyun Woo

AU - Kim, Wan Soo

AU - Lee, Kyogu

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Most of cover song identification algorithms are based on the pairwise (dis)similarity between two songs which are represented by harmonic features such as chroma, and therefore the choice of a distance measure and a feature has a significant impact on performance. Furthermore, since the similarity measure is query-dependent, it cannot represent an absolute distance measure. In this paper, we present a novel approach to tackle the cover song identification problem from a new perspective. We first construct a set of core songs, and represent each song in a high-dimensional space where each dimension indicates the pairwise distance between the given song and the other in the pre-defined core set. There are several advantages to this. First, using a number of reference songs in the core set, we make the most of relative distances to many other songs. Second, as all songs are transformed into the same high-dimensional space, kernel methods and metric learning are exploited for distance computation. Third, our approach does not depend on the computation method for the pairwise distance, and thus can use any existing algorithms. Experimental results confirm that the proposed approach achieved a large performance gain compared to the state-of-the-art methods.

AB - Most of cover song identification algorithms are based on the pairwise (dis)similarity between two songs which are represented by harmonic features such as chroma, and therefore the choice of a distance measure and a feature has a significant impact on performance. Furthermore, since the similarity measure is query-dependent, it cannot represent an absolute distance measure. In this paper, we present a novel approach to tackle the cover song identification problem from a new perspective. We first construct a set of core songs, and represent each song in a high-dimensional space where each dimension indicates the pairwise distance between the given song and the other in the pre-defined core set. There are several advantages to this. First, using a number of reference songs in the core set, we make the most of relative distances to many other songs. Second, as all songs are transformed into the same high-dimensional space, kernel methods and metric learning are exploited for distance computation. Third, our approach does not depend on the computation method for the pairwise distance, and thus can use any existing algorithms. Experimental results confirm that the proposed approach achieved a large performance gain compared to the state-of-the-art methods.

UR - http://www.scopus.com/inward/record.url?scp=85054236711&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054236711&partnerID=8YFLogxK

M3 - Conference contribution

T3 - Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017

SP - 628

EP - 634

BT - Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR 2017

A2 - Duan, Zhiyao

A2 - Turnbull, Douglas

A2 - Hu, Xiao

A2 - Cunningham, Sally Jo

PB - International Society for Music Information Retrieval

ER -