Read-only mirror of the MIREX Wiki, snapshot of 29 October 2026. Open this page on the original wiki to edit it or to see newer changes.

2018:Audio Cover Song Identification Results

From MIREX Wiki

Introduction[edit]

Audio Cover Song (ACS) Identification was evaluated against two datasets: Mixed Collection and Sapp's Mazurka Collection.

Mixed Collection Information[edit]

This is the "original" ACS collection. Within the 1000 pieces in the Audio Cover Song database, there are embedded 30 different "cover songs" each represented by 11 different "versions" for a total of 330 audio files (16bit, monophonic, 22.05khz, wav). The "cover songs" represent a variety of genres (e.g., classical, jazz, gospel, rock, folk-rock, etc.) and the variations span a variety of styles and orchestrations.

Using each of these cover song files in turn as the "seed/query" file, we will examine the returned lists of items for the presence of the other 10 versions of the "seed/query" file.

Sapp's Mazurka Collection Information[edit]

In addition to our original ACS dataset, we used the Mazurka.org dataset put together by Craig Sapp. We randomly chose 11 versions from 49 mazurkas and ran it as a separate ACS subtask. Systems returned a distance matrix of 539x539 from which we located the ranks of each of the associated cover versions.


General Legend[edit]

Sub code Submission name Abstract Contributors
CL1 CL1 PDF Ning Chen, Mingyu Li
LCLL1 CoverNet PDF juheon lee, Sungkyun Chang, Donmoon Lee, Kyogu Lee

Results[edit]

Mixed Collection[edit]

Summary Results[edit]

CL1 LCLL1
Total number of covers identified in top 10 2060.00 2104.00
Mean number of covers identified in top 10 (average performance) 6.24 6.38
Mean (arithmetic) of Avg. Precisions 0.66 0.66
Mean rank of first correctly identified cover 6.89 20.10

download these results as csv

Number of Correct Covers at Rank X Returned in Top Ten[edit]

Rank CL1 LCLL1
1 247 268
2 249 259
3 238 255
4 227 252
5 223 243
6 216 223
7 193 207
8 188 175
9 174 144
10 105 78
Total 2060 2104

download these results as csv

Average Performance per Query Group[edit]

Group CL1 LCLL1
1 0.82 0.82
2 0.13 0.24
3 0.69 0.65
4 0.85 1.00
5 0.78 0.43
6 0.97 0.97
7 0.71 0.70
8 0.49 0.62
9 0.18 0.05
10 0.37 0.52
11 0.28 0.47
12 0.87 0.81
13 0.87 1
14 0.29 0.20
15 0.97 0.97
16 0.74 0.60
17 0.40 0.31
18 0.97 0.94
19 0.71 0.78
20 0.67 0.70
21 0.52 0.63
22 0.73 0.79
23 0.59 0.65
24 0.85 0.83
25 0.37 0.16
26 0.40 0.41
27 0.82 0.82
28 0.78 0.78
29 1 1.00
30 0.90 0.98

download these results as csv

Sapp's Mazuraka Collection[edit]

Summary results[edit]

CL1 LCLL1
Total number of covers identified in top 10 5179.00 5072.00
Mean number of covers identified in top 10 (average performance) 9.61 9.41
Mean (arithmetic) of Avg. Precisions 0.96 0.94
Mean rank of first correctly identified cover 1.36 1.55

download these results as csv

Number of Correct Covers at Rank X Returned in Top Ten[edit]

Rank CL1 LCLL1
1 519 522
2 517 525
3 524 525
4 523 511
5 524 510
6 517 515
7 524 508
8 516 495
9 510 501
10 505 460
Total 5179 5072

download these results as csv

Average Performance per Query Group[edit]

Group CL1 LCLL1
1 1 1
2 1 1
3 1 1
4 1 0.73
5 1 1
6 1 1
7 1 1
8 1 0.99
9 1 0.93
10 1 1
11 1 0.98
12 1 1
13 1 1
14 1 1
15 1 1
16 1 1
17 1 1
18 1 1
19 1 1
20 1 1
21 1 1
22 1 1
23 1 1
24 1 1
25 1 0.98
26 0.59 0.66
27 0.81 0.74
28 0.47 0.45
29 0.35 0.35
30 1 1
31 1 0.97
32 1 1
33 1 0.97
34 1 1
35 1 0.98
36 1 0.91
37 1 1
38 0.94 0.99
39 1 0.93
40 1 1
41 1 1
42 0.88 0.90
43 1 1
44 1 1
45 1 0.94
46 1 1
47 1 1
48 1 1
49 0.83 0.77

download these results as csv

Individual Results Files[edit]

Mixed Collection[edit]

Average Precision by Query

CL1 : Ning Chen, Mingyu Li
LCLL1 : Juheon Lee, Sungkyun Chang, Donmoon Lee, Kyogu Lee

Rank Lists

CL1 : Ning Chen, Mingyu Li
LCLL1 : Juheon Lee, Sungkyun Chang, Donmoon Lee, Kyogu Lee


Mazurkas[edit]

Average Precision by Query

CL1 : Ning Chen, Mingyu Li
LCLL1 : Juheon Lee, Sungkyun Chang, Donmoon Lee, Kyogu Lee


Rank Lists

CL1 : Ning Chen, Mingyu Li
LCLL1 : Juheon Lee, Sungkyun Chang, Donmoon Lee, Kyogu Lee