Table 2 Comparison of matching accuracies using the PV files of the 2009 MIR-QBSH corpus database
From: Intelligent query by humming system based on score level fusion of multiple classifiers
Method | Criterion | ||||
|---|---|---|---|---|---|
| Â | Â | Top 1 (%) | Top 10 (%) | Top 20 (%) | MRR |
LS (Section 2.4.1) | 58.38 | 82.14 | 91.87 | 0.666 | |
Pitch-based DTW (Section 2.4.2) | 56.64 | 84.28 | 92.93 | 0.657 | |
QB code-based LS (Section 2.4.3) | 64.18 | 82.32 | 91.08 | 0.706 | |
Feature point overlap (Section 2.4.4) | 42.16 | 72.85 | 82.66 | 0.525 | |
Distribution (Section 2.4.5) | 12.83 | 57.14 | 78.84 | 0.264 | |
Decision level fusion (OR) | 50.56 | 79.74 | 98.83 | 0.617 | |
Previous method [13] | 70.14 | 86.16 | 93.04 | 0.746 | |
Score level fusion | MIN | 76.74 | 84.39 | 91.53 | 0.789 |
| Â | MAX | 59.42 | 79.92 | 88.08 | 0.667 |
| Â | SUM | 68.27 | 85.93 | 93.41 | 0.743 |
| Â | Weighted SUM | 69.11 | 85.19 | 93.41 | 0.747 |
| Â | PRODUCT (proposed method) | 77.17 | 85.89 | 93.07 | 0.794 |