Two Systems for Automatic Music Genre Recognition: What Are They Really Recognizing?

Bob L. Sturm

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

15 Citations (Scopus)
982 Downloads (Pure)

Abstract

We re-implement and test two state-of-the-art systems
for automatic music genre classification;
but unlike past works in this area,
we look closer than ever before at their behavior.
First, we look at specific instances where
each system consistently applies the same wrong label
across multiple trials of cross-validation.
Second, we test the robustness of each system
to spectral equalization.
Finally, we test how well human subjects recognize
the genres of music excerpts composed by
each system to be highly genre representative.
Our results suggest that neither high-performing system
has a capacity to recognize music genre.
Original languageEnglish
Title of host publicationProceedings of the second international ACM workshop on Music information retrieval with user-centered and multimodal strategies
Volume2012
PublisherAssociation for Computing Machinery
Publication date2012
Pages69-74
ISBN (Print)978-1-4503-1591-3
DOIs
Publication statusPublished - 2012
EventACM Multimedia 2012: Workshop on Music Information Retrieval with User-Centered and Multimodal Strategies - Nara, Japan
Duration: 29 Oct 20122 Nov 2012

Conference

ConferenceACM Multimedia 2012
CountryJapan
CityNara
Period29/10/201202/11/2012
SeriesACM Multimedia

Cite this

Sturm, B. L. (2012). Two Systems for Automatic Music Genre Recognition: What Are They Really Recognizing? In Proceedings of the second international ACM workshop on Music information retrieval with user-centered and multimodal strategies (Vol. 2012, pp. 69-74). Association for Computing Machinery. ACM Multimedia https://doi.org/10.1145/2390848.2390866
Sturm, Bob L. / Two Systems for Automatic Music Genre Recognition : What Are They Really Recognizing?. Proceedings of the second international ACM workshop on Music information retrieval with user-centered and multimodal strategies. Vol. 2012 Association for Computing Machinery, 2012. pp. 69-74 (ACM Multimedia).
@inproceedings{ea3c068e07b34924829036fc1bcb14fc,
title = "Two Systems for Automatic Music Genre Recognition: What Are They Really Recognizing?",
abstract = "We re-implement and test two state-of-the-art systems for automatic music genre classification;but unlike past works in this area, we look closer than ever before at their behavior.First, we look at specific instances whereeach system consistently applies the same wrong labelacross multiple trials of cross-validation.Second, we test the robustness of each systemto spectral equalization.Finally, we test how well human subjects recognizethe genres of music excerpts composed by each system to be highly genre representative.Our results suggest that neither high-performing system has a capacity to recognize music genre.",
author = "Sturm, {Bob L.}",
year = "2012",
doi = "10.1145/2390848.2390866",
language = "English",
isbn = "978-1-4503-1591-3",
volume = "2012",
series = "ACM Multimedia",
pages = "69--74",
booktitle = "Proceedings of the second international ACM workshop on Music information retrieval with user-centered and multimodal strategies",
publisher = "Association for Computing Machinery",
address = "United States",

}

Sturm, BL 2012, Two Systems for Automatic Music Genre Recognition: What Are They Really Recognizing? in Proceedings of the second international ACM workshop on Music information retrieval with user-centered and multimodal strategies. vol. 2012, Association for Computing Machinery, ACM Multimedia, pp. 69-74, ACM Multimedia 2012, Nara, Japan, 29/10/2012. https://doi.org/10.1145/2390848.2390866

Two Systems for Automatic Music Genre Recognition : What Are They Really Recognizing? / Sturm, Bob L.

Proceedings of the second international ACM workshop on Music information retrieval with user-centered and multimodal strategies. Vol. 2012 Association for Computing Machinery, 2012. p. 69-74 (ACM Multimedia).

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

TY - GEN

T1 - Two Systems for Automatic Music Genre Recognition

T2 - What Are They Really Recognizing?

AU - Sturm, Bob L.

PY - 2012

Y1 - 2012

N2 - We re-implement and test two state-of-the-art systems for automatic music genre classification;but unlike past works in this area, we look closer than ever before at their behavior.First, we look at specific instances whereeach system consistently applies the same wrong labelacross multiple trials of cross-validation.Second, we test the robustness of each systemto spectral equalization.Finally, we test how well human subjects recognizethe genres of music excerpts composed by each system to be highly genre representative.Our results suggest that neither high-performing system has a capacity to recognize music genre.

AB - We re-implement and test two state-of-the-art systems for automatic music genre classification;but unlike past works in this area, we look closer than ever before at their behavior.First, we look at specific instances whereeach system consistently applies the same wrong labelacross multiple trials of cross-validation.Second, we test the robustness of each systemto spectral equalization.Finally, we test how well human subjects recognizethe genres of music excerpts composed by each system to be highly genre representative.Our results suggest that neither high-performing system has a capacity to recognize music genre.

UR - http://www.scopus.com/inward/record.url?scp=84870572075&partnerID=8YFLogxK

U2 - 10.1145/2390848.2390866

DO - 10.1145/2390848.2390866

M3 - Article in proceeding

SN - 978-1-4503-1591-3

VL - 2012

T3 - ACM Multimedia

SP - 69

EP - 74

BT - Proceedings of the second international ACM workshop on Music information retrieval with user-centered and multimodal strategies

PB - Association for Computing Machinery

ER -

Sturm BL. Two Systems for Automatic Music Genre Recognition: What Are They Really Recognizing? In Proceedings of the second international ACM workshop on Music information retrieval with user-centered and multimodal strategies. Vol. 2012. Association for Computing Machinery. 2012. p. 69-74. (ACM Multimedia). https://doi.org/10.1145/2390848.2390866