Skip to main content

Table 1 Comparison of the matching accuracies using the PV files of the 2006 MIREX QBSH corpus database

From: Intelligent query by humming system based on score level fusion of multiple classifiers

Method

Criterion

  

Top 1 (%)

Top 10 (%)

Top 20 (%)

MRR

LS (Section 2.4.1)

60.17

82.09

91.96

0.678

Pitch based DTW (Section 2.4.2)

57.81

83.77

92.63

0.663

QB code-based LS (Section 2.4.3)

65.00

83.20

93.78

0.716

Feature point overlap (Section 2.4.4)

41.20

72.45

82.87

0.518

Distribution (Section 2.4.5)

13.09

57.63

78.87

0.264

Decision level fusion (OR)

50.41

80.11

99.00

0.616

Previous method [13]

71.18

86.13

93.17

0.753

Score level fusion

MIN

77.82

85.02

91.62

0.798

 

MAX

58.51

81.12

88.46

0.665

 

SUM

68.72

85.56

92.94

0.745

 

Weighted SUM

69.22

85.60

92.76

0.748

 

PRODUCT (proposed method)

78.14

85.81

93.08

0.802