Difference between revisions of "2009:Music Structure Segmentation Results"

From MIREX Wiki
(Team ID)
(Individual Participant Results)
 
(29 intermediate revisions by 6 users not shown)
Line 1: Line 1:
 +
==Introduction==
 +
This task concerns itself with analyzing the structure of music audio files, and labeling the corresponding segments, e.g. {verse, chorus, bridge, etc}, {A, B, C, etc.}. A more detailed description can be found at the task page [[2009:Structural_Segmentation]]. The dataset consists of 297 popular music songs.
 +
 
===General Legend===
 
===General Legend===
 
====Team ID====
 
====Team ID====
'''EP'''
 
  
'''PK'''
+
'''ANO1''' = [https://www.music-ir.org/mirex/abstracts/2009/Ano.pdf Anonymous] <br />
 +
'''ANO2''' = [https://www.music-ir.org/mirex/abstracts/2009/Ano.pdf Anonymous] <br />
 +
'''MND''' = [https://www.music-ir.org/mirex/abstracts/2009/ACD_SS_mauch.pdf Matthias Mauch, Katy Noland, Simon Dixon] <br />
 +
'''PK''' = [https://www.music-ir.org/mirex/abstracts/2009/PK.pdf Jouni Paulus, Anssi Klapuri]<br />
 +
'''GP''' = [https://www.music-ir.org/mirex/abstracts/2009/Peeters_2009_MIREX_structure.pdf Geoffroy Peeters ] <br />
 +
 
 +
====Evaluation Measures====
 +
'''overSegScore''' - normalised conditional entropy based over-segmentation score, S_o from [http://ismir2008.ismir.net/papers/ISMIR2008_219.pdf Lukashevich ISMIR2008]<br>
 +
'''underSegScore''' - normalised conditional entropy based under-segmentation score, S_u from [http://ismir2008.ismir.net/papers/ISMIR2008_219.pdf Lukashevich ISMIR2008]<br>
 +
'''pwF''' - frame pair clustering F-measure from [http://dx.doi.org/10.1109/TASL.2007.910781 Levy & Sandler TASLP2008]<br>
 +
'''pwPrecision''' - frame pair clustering precision rate from [http://dx.doi.org/10.1109/TASL.2007.910781 Levy & Sandler TASLP2008]<br>
 +
'''pwRecall''' - frame pair clustering recall rate from [http://dx.doi.org/10.1109/TASL.2007.910781 Levy & Sandler TASLP2008]<br>
 +
'''R''' - Rand clustering index from [http://www.springerlink.com/content/x64124718341j1j0/fulltext.pdf Hubert & Arabie, "Comparing partitions", Journal of Classification, 1985]<br>
 +
'''Fmeasure@[0.5, 3]s''' - segment boundary recovery evaluation measure. claimed boundary is accepted if it is within the specified window length from a true boundary, overall F-measure for boundary recovery<br>
 +
'''precRate@[0.5, 3]s''' - segment boundary recovery precision rate<br>
 +
'''recRate@[0.5, 3]s''' - segment boundary recovery recall rate<br>
 +
'''medianTrue2claim''' - median distance from an annotated segment boundary to the closest found boundary, seconds<br>
 +
'''medianClaim2true''' - median distance from a found segment boundary to the closest annotated one, seconds<br>
 +
 
 +
The calculation of the measures is described in [[2009:Structural_Segmentation#Evaluation_Measures]].
 +
 
 +
===MIREX 2009 Music Structure Summary Results - Mean of all Measures===
 +
 
 +
<csv>2009/structure/structure.summary.csv</csv>
 +
 
 +
===MIREX 2009 Music Structure Summary Runtime Data===
 +
<csv>2009/structure/structure.runtime.csv</csv>
  
'''GP'''
+
===Individual Participant Results===
 +
*[[2009:Music_Structure_Segmentation_Results:_AN01]]
 +
*[[2009:Music_Structure_Segmentation_Results:_AN02]]
 +
*[[2009:Music_Structure_Segmentation_Results:_GP]]
 +
*[[2009:Music_Structure_Segmentation_Results:_MND]]
 +
*[[2009:Music_Structure_Segmentation_Results:_PK]]

Latest revision as of 14:41, 1 June 2010

Introduction

This task concerns itself with analyzing the structure of music audio files, and labeling the corresponding segments, e.g. {verse, chorus, bridge, etc}, {A, B, C, etc.}. A more detailed description can be found at the task page 2009:Structural_Segmentation. The dataset consists of 297 popular music songs.

General Legend

Team ID

ANO1 = Anonymous
ANO2 = Anonymous
MND = Matthias Mauch, Katy Noland, Simon Dixon
PK = Jouni Paulus, Anssi Klapuri
GP = Geoffroy Peeters

Evaluation Measures

overSegScore - normalised conditional entropy based over-segmentation score, S_o from Lukashevich ISMIR2008
underSegScore - normalised conditional entropy based under-segmentation score, S_u from Lukashevich ISMIR2008
pwF - frame pair clustering F-measure from Levy & Sandler TASLP2008
pwPrecision - frame pair clustering precision rate from Levy & Sandler TASLP2008
pwRecall - frame pair clustering recall rate from Levy & Sandler TASLP2008
R - Rand clustering index from Hubert & Arabie, "Comparing partitions", Journal of Classification, 1985
Fmeasure@[0.5, 3]s - segment boundary recovery evaluation measure. claimed boundary is accepted if it is within the specified window length from a true boundary, overall F-measure for boundary recovery
precRate@[0.5, 3]s - segment boundary recovery precision rate
recRate@[0.5, 3]s - segment boundary recovery recall rate
medianTrue2claim - median distance from an annotated segment boundary to the closest found boundary, seconds
medianClaim2true - median distance from a found segment boundary to the closest annotated one, seconds

The calculation of the measures is described in 2009:Structural_Segmentation#Evaluation_Measures.

MIREX 2009 Music Structure Summary Results - Mean of all Measures

EvalMeasure ANO1 ANO2 GP MND PK
overSegScore 0.6371 0.65438 0.60119 0.7389 0.59335
underSegScore 0.63701 0.57517 0.67681 0.61756 0.78957
pwF 0.58183 0.57653 0.53254 0.59988 0.54032
pwPrecision 0.59655 0.54254 0.62689 0.56112 0.74148
pwRecall 0.61363 0.67024 0.50493 0.71034 0.46172
R 0.76223 0.73546 0.75931 0.74778 0.79232
Fmeasure@0.5s 0.18312 0.12762 0.1836 0.20951 0.27055
precRate@0.5s 0.16013 0.12538 0.1458 0.15754 0.2429
recRate@0.5s 0.22211 0.13462 0.25972 0.35956 0.32318
Fmeasure@3s 0.5896 0.58447 0.49952 0.39911 0.5305
precRate@3s 0.51527 0.57485 0.39969 0.29931 0.47665
recRate@3s 0.71422 0.61539 0.69782 0.69249 0.63155
medianTrue2claim 1.6264 2.4716 2.1145 2.2301 2.4354
medianClaim2true 3.5477 3.2091 3.52 3.4404 3.5089

download these results as csv

MIREX 2009 Music Structure Summary Runtime Data

Participant Machine Runtime(hh:mm)
AN01 ALE 01:42
ANO2 ALE 01:38
GP CWIN 04:12
MND ALE 09:12
PK ALE 09:53

download these results as csv

Individual Participant Results