Table 4 Comparison of matching accuracies using the extracted pitch data by the STA pitch extractor with 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) | 57.77 | 81.63 | 92.10 | 0.660 | |
Pitch-based DTW (Section 2.4.2) | 55.41 | 83.68 | 92.55 | 0.647 | |
QB code-based LS (Section 2.4.3) | 62.88 | 81.97 | 90.90 | 0.699 | |
Feature point overlap (Section 2.4.4) | 42.00 | 72.74 | 82.69 | 0.527 | |
Distribution (Section 2.4.5) | 12.16 | 58.32 | 79.87 | 0.261 | |
Decision level fusion (OR) | 51.28 | 79.30 | 99.07 | 0.619 | |
Previous method [13] | 69.10 | 85.42 | 92.78 | 0.736 | |
Score level fusion | MIN | 77.12 | 83.93 | 91.29 | 0.790 |
| Â | MAX | 59.13 | 81.09 | 88.29 | 0.665 |
| Â | SUM | 68.97 | 85.71 | 93.55 | 0.745 |
| Â | Weighted SUM | 69.08 | 85.20 | 93.41 | 0.745 |
| Â | PRODUCT (proposed method) | 77.27 | 85.56 | 93.12 | 0.793 |