¿El Caballo Viejo? Latin Genre Recognition with Deep Learning and Spectral Periodicity

Bob L. Sturm; Corey Mose Kereliuk; Jan Larsen

doi:10.1007/978-3-319-20603-5_34

¿El Caballo Viejo? Latin Genre Recognition with Deep Learning and Spectral Periodicity

Bob L. Sturm, Corey Mose Kereliuk, Jan Larsen

Queen Mary University of London

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Abstract

The “winning” system in the 2013 MIREX Latin Genre Classification Task was a deep neural network trained with simple features. An explanation for its winning performance has yet to be found. In previous work, we built similar systems using the BALLROOM music dataset, and found their performances to be greatly affected by slightly changing the tempo of the music of a test recording. In the MIREX task, however, systems are trained and tested using the Latin Music Dataset (LMD), which is 4.5 times larger than BALLROOM, and which does not seem to show as strong a relationship between tempo and label as BALLROOM. In this paper, we reproduce the “winning” deep learning system using LMD, and measure the effects of time dilation on its performance. We find that tempo changes of at most ±6 % greatly diminish and improve its performance. Interpreted with the low-level nature of the input features, this supports the conclusion that the system is exploiting some low-level absolute time characteristics to reproduce ground truth in LMD.

Original language	English
Title of host publication	Proceedings of the 5th International Conference Mathematics and Computation in Music, MCM 2015
Editors	Tom Collins, David Meredith, Anja Volk
Publisher	Springer
Publication date	2015
Pages	335-346
ISBN (Print)	978-3-319-20602-8
ISBN (Electronic)	978-3-319-20603-5
DOIs	https://doi.org/10.1007/978-3-319-20603-5_34
Publication status	Published - 2015
Event	5th International Conference on Mathematics and Computation in Music (MCM 2015) - London, United Kingdom Duration: 22 Jun 2015 → 25 Jun 2015 Conference number: 5 http://mcm2015.qmul.ac.uk/

Conference

Conference	5th International Conference on Mathematics and Computation in Music (MCM 2015)
Number	5
Country/Territory	United Kingdom
City	London
Period	22/06/2015 → 25/06/2015
Internet address	http://mcm2015.qmul.ac.uk/

Series	Lecture Notes in Computer Science
Volume	9110
ISSN	0302-9743

Keywords

Machine music listening
Genre
Deep Learning
Evaluation

Access to Document

10.1007/978-3-319-20603-5_34

OpenUrl availability

Full text

Cite this

@inproceedings{fd05b7b2e6f14306b8b7cd41b3753137,

title = "¿El Caballo Viejo? Latin Genre Recognition with Deep Learning and Spectral Periodicity",

abstract = "The “winning” system in the 2013 MIREX Latin Genre Classification Task was a deep neural network trained with simple features. An explanation for its winning performance has yet to be found. In previous work, we built similar systems using the BALLROOM music dataset, and found their performances to be greatly affected by slightly changing the tempo of the music of a test recording. In the MIREX task, however, systems are trained and tested using the Latin Music Dataset (LMD), which is 4.5 times larger than BALLROOM, and which does not seem to show as strong a relationship between tempo and label as BALLROOM. In this paper, we reproduce the “winning” deep learning system using LMD, and measure the effects of time dilation on its performance. We find that tempo changes of at most ±6 % greatly diminish and improve its performance. Interpreted with the low-level nature of the input features, this supports the conclusion that the system is exploiting some low-level absolute time characteristics to reproduce ground truth in LMD.",

keywords = "Machine music listening, Genre, Deep Learning, Evaluation",

author = "Sturm, {Bob L.} and Kereliuk, {Corey Mose} and Jan Larsen",

year = "2015",

doi = "10.1007/978-3-319-20603-5_34",

language = "English",

isbn = "978-3-319-20602-8",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "335--346",

editor = "Collins, {Tom } and Meredith, {David } and Volk, {Anja }",

booktitle = "Proceedings of the 5th International Conference Mathematics and Computation in Music, MCM 2015",

note = "5th International Conference on Mathematics and Computation in Music (MCM 2015), MCM ; Conference date: 22-06-2015 Through 25-06-2015",

url = "http://mcm2015.qmul.ac.uk/",

}

Sturm, BL, Kereliuk, CM & Larsen, J 2015, ¿El Caballo Viejo? Latin Genre Recognition with Deep Learning and Spectral Periodicity. in T Collins, D Meredith & A Volk (eds), Proceedings of the 5th International Conference Mathematics and Computation in Music, MCM 2015. Springer, Lecture Notes in Computer Science, vol. 9110, pp. 335-346, 5th International Conference on Mathematics and Computation in Music (MCM 2015), London, United Kingdom, 22/06/2015. https://doi.org/10.1007/978-3-319-20603-5_34

¿El Caballo Viejo? Latin Genre Recognition with Deep Learning and Spectral Periodicity. / Sturm, Bob L.; Kereliuk, Corey Mose; Larsen, Jan.
Proceedings of the 5th International Conference Mathematics and Computation in Music, MCM 2015. ed. / Tom Collins; David Meredith; Anja Volk. Springer, 2015. p. 335-346 (Lecture Notes in Computer Science, Vol. 9110).

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - ¿El Caballo Viejo? Latin Genre Recognition with Deep Learning and Spectral Periodicity

AU - Sturm, Bob L.

AU - Kereliuk, Corey Mose

AU - Larsen, Jan

N1 - Conference code: 5

PY - 2015

Y1 - 2015

N2 - The “winning” system in the 2013 MIREX Latin Genre Classification Task was a deep neural network trained with simple features. An explanation for its winning performance has yet to be found. In previous work, we built similar systems using the BALLROOM music dataset, and found their performances to be greatly affected by slightly changing the tempo of the music of a test recording. In the MIREX task, however, systems are trained and tested using the Latin Music Dataset (LMD), which is 4.5 times larger than BALLROOM, and which does not seem to show as strong a relationship between tempo and label as BALLROOM. In this paper, we reproduce the “winning” deep learning system using LMD, and measure the effects of time dilation on its performance. We find that tempo changes of at most ±6 % greatly diminish and improve its performance. Interpreted with the low-level nature of the input features, this supports the conclusion that the system is exploiting some low-level absolute time characteristics to reproduce ground truth in LMD.

AB - The “winning” system in the 2013 MIREX Latin Genre Classification Task was a deep neural network trained with simple features. An explanation for its winning performance has yet to be found. In previous work, we built similar systems using the BALLROOM music dataset, and found their performances to be greatly affected by slightly changing the tempo of the music of a test recording. In the MIREX task, however, systems are trained and tested using the Latin Music Dataset (LMD), which is 4.5 times larger than BALLROOM, and which does not seem to show as strong a relationship between tempo and label as BALLROOM. In this paper, we reproduce the “winning” deep learning system using LMD, and measure the effects of time dilation on its performance. We find that tempo changes of at most ±6 % greatly diminish and improve its performance. Interpreted with the low-level nature of the input features, this supports the conclusion that the system is exploiting some low-level absolute time characteristics to reproduce ground truth in LMD.

KW - Machine music listening

KW - Genre

KW - Deep Learning

KW - Evaluation

U2 - 10.1007/978-3-319-20603-5_34

DO - 10.1007/978-3-319-20603-5_34

M3 - Article in proceedings

SN - 978-3-319-20602-8

T3 - Lecture Notes in Computer Science

SP - 335

EP - 346

BT - Proceedings of the 5th International Conference Mathematics and Computation in Music, MCM 2015

A2 - Collins, Tom

A2 - Meredith, David

A2 - Volk, Anja

PB - Springer

T2 - 5th International Conference on Mathematics and Computation in Music (MCM 2015)

Y2 - 22 June 2015 through 25 June 2015

ER -

Sturm BL, Kereliuk CM, Larsen J. ¿El Caballo Viejo? Latin Genre Recognition with Deep Learning and Spectral Periodicity. In Collins T, Meredith D, Volk A, editors, Proceedings of the 5th International Conference Mathematics and Computation in Music, MCM 2015. Springer. 2015. p. 335-346. (Lecture Notes in Computer Science, Vol. 9110). doi: 10.1007/978-3-319-20603-5_34

¿El Caballo Viejo? Latin Genre Recognition with Deep Learning and Spectral Periodicity

Abstract

Conference

Keywords

Access to Document

OpenUrl availability

Fingerprint

Cite this