A long-range self-similarity approach to segmenting DJ mixed music streams

Tim Scarfe; Wouter M. Koolen; Yuri Kalnishkan

doi:10.1007/978-3-642-41142-7_24

A long-range self-similarity approach to segmenting DJ mixed music streams

Tim Scarfe
, Wouter M. Koolen
, Yuri Kalnishkan

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

1134 Downloads (Pure)

Abstract

In this paper we describe an unsupervised, deterministic algorithm for segmenting DJ-mixed Electronic Dance Music streams (for example; podcasts, radio shows, live events) into their respective tracks. We attempt to reconstruct boundaries as close as possible to what a human domain expert would engender. The goal of DJ-mixing is to render track boundaries effectively invisible from the standpoint of human perception which makes the problem difficult.

We use Dynamic Programming (DP) to optimally segment a cost matrix derived from a similarity matrix. The similarity matrix is based on the cosines of a time series of kernel-transformed Fourier based features designed with this domain in mind. Our method is applied to EDM streams. Its formulation incorporates long-term self similarity as a first class concept combined with DP and it is qualitatively assessed on a large corpus of long streams that have been hand labelled by a domain expert.

Original language	English
Title of host publication	Artificial Intelligence Applications and Innovations
Subtitle of host publication	Proceedings of the 9th IFIP WG 12.5 International Conference, AIAI 2013, Paphos, Cyprus, September 30 – October 2, 2013
Editors	Harris Papadopoulos, Andreas S. Andreou, Lazaros Iliadis, Ilias Maglogiannis
Publisher	Springer
Pages	235-244
Number of pages	10
ISBN (Electronic)	978-3-642-41142-7
ISBN (Print)	978-3-642-41141-0
DOIs	https://doi.org/10.1007/978-3-642-41142-7_24
Publication status	Published - 2013
Event	9th IFIP International Conference, AIAI 2013 - Paphos, Cyprus Duration: 30 Sept 2013 → 2 Oct 2013

Publication series

Name	IFIP Advances in Information and Communication Technology
Volume	412
ISSN (Print)	1868-4238

Conference

Conference	9th IFIP International Conference, AIAI 2013
Country/Territory	Cyprus
City	Paphos
Period	30/09/13 → 2/10/13

Keywords

Music
SEGMENTATION
DJ mix
dynamic programming

Access to Document

10.1007/978-3-642-41142-7_24

Segmenting DJ Mixed Music Streams
Copyright Holder IFIP International Federation for Information Processing
Submitted manuscript, 302 KB

http://link.springer.com/chapter/10.1007/978-3-642-41142-7_24

Cite this

Scarfe, T., Koolen, W. M., & Kalnishkan, Y. (2013). A long-range self-similarity approach to segmenting DJ mixed music streams. In H. Papadopoulos, A. S. Andreou, L. Iliadis, & I. Maglogiannis (Eds.), Artificial Intelligence Applications and Innovations: Proceedings of the 9th IFIP WG 12.5 International Conference, AIAI 2013, Paphos, Cyprus, September 30 – October 2, 2013 (pp. 235-244). (IFIP Advances in Information and Communication Technology; Vol. 412). Springer. https://doi.org/10.1007/978-3-642-41142-7_24

Scarfe, Tim ; Koolen, Wouter M. ; Kalnishkan, Yuri. / A long-range self-similarity approach to segmenting DJ mixed music streams. Artificial Intelligence Applications and Innovations: Proceedings of the 9th IFIP WG 12.5 International Conference, AIAI 2013, Paphos, Cyprus, September 30 – October 2, 2013. editor / Harris Papadopoulos ; Andreas S. Andreou ; Lazaros Iliadis ; Ilias Maglogiannis. Springer, 2013. pp. 235-244 (IFIP Advances in Information and Communication Technology).

@inproceedings{d4a3bd733a5d4d40a2b5dd575013e2ab,

title = "A long-range self-similarity approach to segmenting DJ mixed music streams",

abstract = "In this paper we describe an unsupervised, deterministic algorithm for segmenting DJ-mixed Electronic Dance Music streams (for example; podcasts, radio shows, live events) into their respective tracks. We attempt to reconstruct boundaries as close as possible to what a human domain expert would engender. The goal of DJ-mixing is to render track boundaries effectively invisible from the standpoint of human perception which makes the problem difficult. We use Dynamic Programming (DP) to optimally segment a cost matrix derived from a similarity matrix. The similarity matrix is based on the cosines of a time series of kernel-transformed Fourier based features designed with this domain in mind. Our method is applied to EDM streams. Its formulation incorporates long-term self similarity as a first class concept combined with DP and it is qualitatively assessed on a large corpus of long streams that have been hand labelled by a domain expert.",

keywords = "Music, SEGMENTATION, DJ mix, dynamic programming",

author = "Tim Scarfe and Koolen, \{Wouter M.\} and Yuri Kalnishkan",

year = "2013",

doi = "10.1007/978-3-642-41142-7\_24",

language = "English",

isbn = "978-3-642-41141-0",

series = "IFIP Advances in Information and Communication Technology",

publisher = "Springer",

pages = "235--244",

editor = "Harris Papadopoulos and Andreou, \{Andreas S.\} and Lazaros Iliadis and Ilias Maglogiannis",

booktitle = "Artificial Intelligence Applications and Innovations",

note = "9th IFIP International Conference, AIAI 2013 ; Conference date: 30-09-2013 Through 02-10-2013",

}

Scarfe, T, Koolen, WM & Kalnishkan, Y 2013, A long-range self-similarity approach to segmenting DJ mixed music streams. in H Papadopoulos, AS Andreou, L Iliadis & I Maglogiannis (eds), Artificial Intelligence Applications and Innovations: Proceedings of the 9th IFIP WG 12.5 International Conference, AIAI 2013, Paphos, Cyprus, September 30 – October 2, 2013. IFIP Advances in Information and Communication Technology, vol. 412, Springer, pp. 235-244, 9th IFIP International Conference, AIAI 2013, Paphos, Cyprus, 30/09/13. https://doi.org/10.1007/978-3-642-41142-7_24

A long-range self-similarity approach to segmenting DJ mixed music streams. / Scarfe, Tim; Koolen, Wouter M.; Kalnishkan, Yuri.
Artificial Intelligence Applications and Innovations: Proceedings of the 9th IFIP WG 12.5 International Conference, AIAI 2013, Paphos, Cyprus, September 30 – October 2, 2013. ed. / Harris Papadopoulos; Andreas S. Andreou; Lazaros Iliadis; Ilias Maglogiannis. Springer, 2013. p. 235-244 (IFIP Advances in Information and Communication Technology; Vol. 412).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - A long-range self-similarity approach to segmenting DJ mixed music streams

AU - Scarfe, Tim

AU - Koolen, Wouter M.

AU - Kalnishkan, Yuri

PY - 2013

Y1 - 2013

N2 - In this paper we describe an unsupervised, deterministic algorithm for segmenting DJ-mixed Electronic Dance Music streams (for example; podcasts, radio shows, live events) into their respective tracks. We attempt to reconstruct boundaries as close as possible to what a human domain expert would engender. The goal of DJ-mixing is to render track boundaries effectively invisible from the standpoint of human perception which makes the problem difficult. We use Dynamic Programming (DP) to optimally segment a cost matrix derived from a similarity matrix. The similarity matrix is based on the cosines of a time series of kernel-transformed Fourier based features designed with this domain in mind. Our method is applied to EDM streams. Its formulation incorporates long-term self similarity as a first class concept combined with DP and it is qualitatively assessed on a large corpus of long streams that have been hand labelled by a domain expert.

AB - In this paper we describe an unsupervised, deterministic algorithm for segmenting DJ-mixed Electronic Dance Music streams (for example; podcasts, radio shows, live events) into their respective tracks. We attempt to reconstruct boundaries as close as possible to what a human domain expert would engender. The goal of DJ-mixing is to render track boundaries effectively invisible from the standpoint of human perception which makes the problem difficult. We use Dynamic Programming (DP) to optimally segment a cost matrix derived from a similarity matrix. The similarity matrix is based on the cosines of a time series of kernel-transformed Fourier based features designed with this domain in mind. Our method is applied to EDM streams. Its formulation incorporates long-term self similarity as a first class concept combined with DP and it is qualitatively assessed on a large corpus of long streams that have been hand labelled by a domain expert.

KW - Music

KW - SEGMENTATION

KW - DJ mix

KW - dynamic programming

U2 - 10.1007/978-3-642-41142-7_24

DO - 10.1007/978-3-642-41142-7_24

M3 - Conference contribution

SN - 978-3-642-41141-0

T3 - IFIP Advances in Information and Communication Technology

SP - 235

EP - 244

BT - Artificial Intelligence Applications and Innovations

A2 - Papadopoulos, Harris

A2 - Andreou, Andreas S.

A2 - Iliadis, Lazaros

A2 - Maglogiannis, Ilias

PB - Springer

T2 - 9th IFIP International Conference, AIAI 2013

Y2 - 30 September 2013 through 2 October 2013

ER -

Scarfe T, Koolen WM, Kalnishkan Y. A long-range self-similarity approach to segmenting DJ mixed music streams. In Papadopoulos H, Andreou AS, Iliadis L, Maglogiannis I, editors, Artificial Intelligence Applications and Innovations: Proceedings of the 9th IFIP WG 12.5 International Conference, AIAI 2013, Paphos, Cyprus, September 30 – October 2, 2013. Springer. 2013. p. 235-244. (IFIP Advances in Information and Communication Technology). doi: 10.1007/978-3-642-41142-7_24