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.

2006:Audio Onset Detection Results

From MIREX Wiki

Introduction[edit]

These are the results for the 2006 running of the Audio Onset Detection task set. For background information about this task set please refer to the 2006:Audio Onset Detection page.

The aim of the Audio Onset Detection task is to find the time locations at which all musical events in a recording begin. The dataset consists of 85 recordings across 9 different "classes" (e.g. solo drums, polyphonic pitched, etc.). For each sound file, ground truth annotations produced by 3-5 listeners were used for the evaluation. Each algorithm was tested across 10-20 different parameterizations (e.g. thresholds) in order to produce Precision vs. Recall Operating Characteristic (P-ROC) curves. The primary evauluation metric used was the F1-Measure (the equal weighted harmonic mean of precision and recall).

General Legend[edit]

Team ID[edit]

dixon = Simon Dixon
roebel = A. Röbel
brossier = Paul Brossier
du = Yunfeng Du, Ming Li, Jian Liu

  • Dixon's NWPD submission was modified by Andreas Ehmann, and requires the author's verification

Overall Summary Results[edit]

MIREX 2006 Audio Onset Detection Summary Results - Peak F-measure performance across all parameterizations[edit]

Contestant Parameters Total Correct Total FP Total FN Total Merged Total Doubled Avg. Correct Avg. FP Avg. FN Avg. Merged Avg. Doubled Avg. Precision Avg. Recall Avg. F-Measure
brossier_complex 0.45 6407 1709 3092 133 387 22.169 6.067 9.5 0.429 1.3 0.78 0.725 0.721
brossier_dual 0.4 6930 1979 2569 109 869 23.271 6.459 8.398 0.347 2.777 0.769 0.735 0.724
brossier_hfc 0.25 7368 2573 2131 115 884 24.645 8.402 7.024 0.358 2.706 0.752 0.774 0.734
brossier_specdiff 0.4 6475 1757 3024 126 481 21.963 5.731 9.705 0.394 1.515 0.764 0.701 0.707
dixon_cd (0.85/ 0.30) 6945 3948 2554 172 120 23.94 13.319 7.729 0.536 0.408 0.709 0.776 0.71
dixon_nwpd (0.89/ 0.60) 8460 10431 1039 176 820 28.522 35.842 3.146 0.551 2.693 0.524 0.908 0.62
dixon_rcd (0.88/ 0.70) 6867 3014 2632 161 167 23.598 10.202 8.071 0.492 0.591 0.735 0.765 0.716
dixon_sf (0.90/ 0.55) 7217 3684 2282 176 159 24.655 12.369 7.013 0.538 0.536 0.736 0.79 0.726
dixon_wpd (0.89/ 0.65) 7435 4820 2064 187 335 25.204 15.749 6.464 0.576 1.071 0.663 0.786 0.685
du 1.3 7353 3142 2146 193 7 24.697 10.491 6.971 0.597 0.023 0.797 0.799 0.762
roebel_1 0.09 6795 1005 2704 192 28 23.104 3.367 8.565 0.591 0.086 0.861 0.746 0.777
roebel_2 0.09 7048 1282 2451 201 50 23.885 4.279 7.783 0.619 0.146 0.831 0.769 0.78
roebel_3 0.06 7173 1323 2326 200 51 24.321 4.385 7.347 0.612 0.153 0.836 0.779 0.788

download these results as csv

MIREX 2006 Audio Onset Detection Summary Plot[edit]

File:Onset06 summary.png

MIREX 2006 Audio Onset Detection Runtime Data[edit]

Contestant Machine Avg. run time per parameter set
brossier LINUX 34
dixon FAST 966
du FAST 64
roebel LINUX 327

download these results as csv

Results by Class[edit]

Individual Results[edit]