Difference between revisions of "2006:Audio Beat Tracking Results"
Kahyun Choi (talk | contribs) (→Summary, plots, and comments on the results) |
|||
| (9 intermediate revisions by 3 users not shown) | |||
| Line 1: | Line 1: | ||
[[Category: Results]] | [[Category: Results]] | ||
==Introduction== | ==Introduction== | ||
| − | These are the results for the 2006 running of the Audio Beat Tracking task set. For background information about this task set please refer to the [[Audio Beat Tracking]] page. | + | These are the results for the 2006 running of the Audio Beat Tracking task set. For background information about this task set please refer to the [[2006:Audio Beat Tracking]] page. |
| + | |||
| + | '''NOTE:''' Due to an evaluation error, the results of the Audio Beat Tracking task have been updated as of 17 July 2007, and differ from those presented at ISMIR 2006. | ||
The aim of the automatic beat tracking task is to track each beat location in a collection of 140 sound files. Unlike the Audio Tempo Extraction task, which aims to detect tempi for each file, the beat tracking task aims at detecting all beat locations in the recordings. The algorithms were evaluated in terms of their accuracy in predicting beat locations annotated by a group of roughly 40 listeners per sound file. Because listeners may tap at different rates (e.g. at twice the rate of another annotator, or other multiples for non-binary meters), it is difficult to assess what a 'perfect' score may be. | The aim of the automatic beat tracking task is to track each beat location in a collection of 140 sound files. Unlike the Audio Tempo Extraction task, which aims to detect tempi for each file, the beat tracking task aims at detecting all beat locations in the recordings. The algorithms were evaluated in terms of their accuracy in predicting beat locations annotated by a group of roughly 40 listeners per sound file. Because listeners may tap at different rates (e.g. at twice the rate of another annotator, or other multiples for non-binary meters), it is difficult to assess what a 'perfect' score may be. | ||
| Line 8: | Line 10: | ||
====Team ID==== | ====Team ID==== | ||
| − | '''dixon''' = [https://www.music-ir.org/ | + | '''dixon''' = [https://www.music-ir.org/mirex/abstracts/2006/BT_dixon.pdf Simon Dixon]<br /> |
| − | '''ellis''' = [https://www.music-ir.org/ | + | '''ellis''' = [https://www.music-ir.org/mirex/abstracts/2006/TE_BT_ellis.pdf Daniel P. W. Ellis]<br /> |
| − | '''klapuri''' = [https://www.music-ir.org/ | + | '''klapuri''' = [https://www.music-ir.org/mirex/abstracts/2006/BT_klapuri.pdf Anssi Klapuri]<br /> |
| − | '''davies''' = [https://www.music-ir.org/ | + | '''davies''' = [https://www.music-ir.org/mirex/abstracts/2006/BT_TE_davies.pdf M. E. P. Davies and M. D. Plumbley]<br /> |
| − | '''brossier''' = [https://www.music-ir.org/ | + | '''brossier''' = [https://www.music-ir.org/mirex/abstracts/2006/AME_BT_OD_TE_brossier.pdf Paul Brossier]<br /> |
==Overall Summary Results== | ==Overall Summary Results== | ||
===MIREX 2006 Audio Beat Tracking Summary Results=== | ===MIREX 2006 Audio Beat Tracking Summary Results=== | ||
| − | <csv>ab06_results.csv</csv> | + | <csv>2006/ab06_results.csv</csv> |
===MIREX 2006 Audio Beat Tracking Runtime Data=== | ===MIREX 2006 Audio Beat Tracking Runtime Data=== | ||
| − | <csv>ab06_runtime.csv</csv> | + | <csv>2006/ab06_runtime.csv</csv> |
===Summary, plots, and comments on the results=== | ===Summary, plots, and comments on the results=== | ||
| − | + | 2006:Media:Beat_and_tempo_results.pdf<!--Click here to get a summary, plots and comments on the contest.!--> | |
Latest revision as of 22:46, 19 December 2011
Contents
Introduction
These are the results for the 2006 running of the Audio Beat Tracking task set. For background information about this task set please refer to the 2006:Audio Beat Tracking page.
NOTE: Due to an evaluation error, the results of the Audio Beat Tracking task have been updated as of 17 July 2007, and differ from those presented at ISMIR 2006.
The aim of the automatic beat tracking task is to track each beat location in a collection of 140 sound files. Unlike the Audio Tempo Extraction task, which aims to detect tempi for each file, the beat tracking task aims at detecting all beat locations in the recordings. The algorithms were evaluated in terms of their accuracy in predicting beat locations annotated by a group of roughly 40 listeners per sound file. Because listeners may tap at different rates (e.g. at twice the rate of another annotator, or other multiples for non-binary meters), it is difficult to assess what a 'perfect' score may be.
General Legend
Team ID
dixon = Simon Dixon
ellis = Daniel P. W. Ellis
klapuri = Anssi Klapuri
davies = M. E. P. Davies and M. D. Plumbley
brossier = Paul Brossier
Overall Summary Results
MIREX 2006 Audio Beat Tracking Summary Results
| Contestant | P-Score (average) |
|---|---|
| dixon | 0.575 |
| davies | 0.571 |
| klapuri | 0.564 |
| ellis | 0.552 |
| brossier | 0.453 |
MIREX 2006 Audio Beat Tracking Runtime Data
| Contestant | Machine | Run-time(seconds) |
|---|---|---|
| brossier | LINUX | 139 |
| davies | FAST | 1394 |
| dixon | FAST | 639 |
| ellis | LINUX | 498 |
| klapuri | LINUX | 1218 |
Summary, plots, and comments on the results
2006:Media:Beat_and_tempo_results.pdf
