Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations

Hendrik Vincent Koops; W.B. de Haas; J. Bransen; A. Volk

doi:10.13140/RG.2.2.22227.99364/1

Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations

Hendrik Vincent Koops, W.B. de Haas, J. Bransen, A. Volk

Chordify

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

Abstract

The increasing accuracy of automatic chord estimation systems, the availability of vast amounts of heterogeneous reference annotations, and insights from annotator subjectivity research make chord label personalization increasingly important. Nevertheless, automatic chord estimation systems are historically exclusively trained and evaluated on a single reference annotation. We introduce a first approach to automatic chord label personalization by modeling subjectivity through deep learning of a harmonic interval-based chord label representation. After integrating these representations from multiple annotators, we can accurately personalize chord labels for individual annotators from a single model and the annotators’ chord label vocabulary. Furthermore, we show that chord personalization using multiple reference annotations outperforms using a single reference annotation.

Original language	English
Title of host publication	Proceedings of the first International Workshop on Deep Learning and Music
Editors	Dorien Herremans, Ching-Hua Chuan
Place of Publication	Anchorage, Alaska, USA
Pages	19-25
Number of pages	7
DOIs	https://doi.org/10.13140/RG.2.2.22227.99364/1
Publication status	Published - 18 May 2017
Event	International Workshop on Deep Learning and Music - William A. Egan Civic and Convention Center, Anchorage, United States Duration: 18 May 2017 → 19 May 2017

Publication series

Name	Proceedings of the International Workshop on Deep Learning and Music
Volume	1

Workshop

Workshop	International Workshop on Deep Learning and Music
Abbreviated title	DLM2017
Country/Territory	United States
City	Anchorage
Period	18/05/17 → 19/05/17

Keywords

Automatic Chord Estimation
Annotator Subjectivity
Deep Learning

Access to Document

10.13140/RG.2.2.22227.99364/1

Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based RepresentationsFinal published version, 355 KB

Cite this

Koops, H. V., de Haas, W. B., Bransen, J., & Volk, A. (2017). Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations. In D. Herremans, & C.-H. Chuan (Eds.), Proceedings of the first International Workshop on Deep Learning and Music (pp. 19-25). (Proceedings of the International Workshop on Deep Learning and Music; Vol. 1).. https://doi.org/10.13140/RG.2.2.22227.99364/1

Koops, Hendrik Vincent ; de Haas, W.B. ; Bransen, J. et al. / Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations. Proceedings of the first International Workshop on Deep Learning and Music. editor / Dorien Herremans ; Ching-Hua Chuan. Anchorage, Alaska, USA, 2017. pp. 19-25 (Proceedings of the International Workshop on Deep Learning and Music).

@inproceedings{f8434a8fd3944abbaf0b6145746b0375,

title = "Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations",

abstract = "The increasing accuracy of automatic chord estimation systems, the availability of vast amounts of heterogeneous reference annotations, and insights from annotator subjectivity research make chord label personalization increasingly important. Nevertheless, automatic chord estimation systems are historically exclusively trained and evaluated on a single reference annotation. We introduce a first approach to automatic chord label personalization by modeling subjectivity through deep learning of a harmonic interval-based chord label representation. After integrating these representations from multiple annotators, we can accurately personalize chord labels for individual annotators from a single model and the annotators{\textquoteright} chord label vocabulary. Furthermore, we show that chord personalization using multiple reference annotations outperforms using a single reference annotation.",

keywords = "Automatic Chord Estimation, Annotator Subjectivity, Deep Learning",

author = "Koops, \{Hendrik Vincent\} and \{de Haas\}, W.B. and J. Bransen and A. Volk",

year = "2017",

month = may,

day = "18",

doi = "10.13140/RG.2.2.22227.99364/1",

language = "English",

series = "Proceedings of the International Workshop on Deep Learning and Music",

pages = "19--25",

editor = "Dorien Herremans and Ching-Hua Chuan",

booktitle = "Proceedings of the first International Workshop on Deep Learning and Music",

note = "International Workshop on Deep Learning and Music, DLM2017 ; Conference date: 18-05-2017 Through 19-05-2017",

}

Koops, HV, de Haas, WB, Bransen, J & Volk, A 2017, Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations. in D Herremans & C-H Chuan (eds), Proceedings of the first International Workshop on Deep Learning and Music. Proceedings of the International Workshop on Deep Learning and Music, vol. 1, Anchorage, Alaska, USA, pp. 19-25, International Workshop on Deep Learning and Music, Anchorage, United States, 18/05/17. https://doi.org/10.13140/RG.2.2.22227.99364/1

Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations. / Koops, Hendrik Vincent; de Haas, W.B.; Bransen, J. et al.
Proceedings of the first International Workshop on Deep Learning and Music. ed. / Dorien Herremans; Ching-Hua Chuan. Anchorage, Alaska, USA, 2017. p. 19-25 (Proceedings of the International Workshop on Deep Learning and Music; Vol. 1).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations

AU - Koops, Hendrik Vincent

AU - de Haas, W.B.

AU - Bransen, J.

AU - Volk, A.

PY - 2017/5/18

Y1 - 2017/5/18

N2 - The increasing accuracy of automatic chord estimation systems, the availability of vast amounts of heterogeneous reference annotations, and insights from annotator subjectivity research make chord label personalization increasingly important. Nevertheless, automatic chord estimation systems are historically exclusively trained and evaluated on a single reference annotation. We introduce a first approach to automatic chord label personalization by modeling subjectivity through deep learning of a harmonic interval-based chord label representation. After integrating these representations from multiple annotators, we can accurately personalize chord labels for individual annotators from a single model and the annotators’ chord label vocabulary. Furthermore, we show that chord personalization using multiple reference annotations outperforms using a single reference annotation.

AB - The increasing accuracy of automatic chord estimation systems, the availability of vast amounts of heterogeneous reference annotations, and insights from annotator subjectivity research make chord label personalization increasingly important. Nevertheless, automatic chord estimation systems are historically exclusively trained and evaluated on a single reference annotation. We introduce a first approach to automatic chord label personalization by modeling subjectivity through deep learning of a harmonic interval-based chord label representation. After integrating these representations from multiple annotators, we can accurately personalize chord labels for individual annotators from a single model and the annotators’ chord label vocabulary. Furthermore, we show that chord personalization using multiple reference annotations outperforms using a single reference annotation.

KW - Automatic Chord Estimation

KW - Annotator Subjectivity

KW - Deep Learning

U2 - 10.13140/RG.2.2.22227.99364/1

DO - 10.13140/RG.2.2.22227.99364/1

M3 - Conference contribution

T3 - Proceedings of the International Workshop on Deep Learning and Music

SP - 19

EP - 25

BT - Proceedings of the first International Workshop on Deep Learning and Music

A2 - Herremans, Dorien

A2 - Chuan, Ching-Hua

CY - Anchorage, Alaska, USA

T2 - International Workshop on Deep Learning and Music

Y2 - 18 May 2017 through 19 May 2017

ER -

Koops HV, de Haas WB, Bransen J, Volk A. Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations. In Herremans D, Chuan CH, editors, Proceedings of the first International Workshop on Deep Learning and Music. Anchorage, Alaska, USA. 2017. p. 19-25. (Proceedings of the International Workshop on Deep Learning and Music). doi: 10.13140/RG.2.2.22227.99364/1

Chord Label Personalization through Deep Learning of Integrated Harmonic Interval-based Representations

Abstract

Publication series

Workshop

Keywords

Access to Document

Fingerprint

Cite this