README.md 3.21 KB
Newer Older
Sebastian Böck's avatar
Sebastian Böck committed
1
# Onset dataset
Sebastian Böck's avatar
Sebastian Böck committed
2

Sebastian Böck's avatar
Sebastian Böck committed
3
This dataset is described in [1][1].
Sebastian Böck's avatar
Sebastian Böck committed
4

Sebastian Böck's avatar
Sebastian Böck committed
5
## Contact:
Sebastian Böck's avatar
Sebastian Böck committed
6

Sebastian Böck's avatar
Sebastian Böck committed
7
Sebastian Böck <sebastian.boeck@jku.at>
Sebastian Böck's avatar
Sebastian Böck committed
8

Sebastian Böck's avatar
Sebastian Böck committed
9
## Description:
Sebastian Böck's avatar
Sebastian Böck committed
10

Sebastian Böck's avatar
Sebastian Böck committed
11
Collection of 321 audio excerpts with onset annotations.
Sebastian Böck's avatar
Sebastian Böck committed
12

Sebastian Böck's avatar
Sebastian Böck committed
13
14
15
## Content:

```
Sebastian Böck's avatar
Sebastian Böck committed
16
17
18
19
20
audio/                  original audio files converted to .flac format
annotations/onsets/     onset annotations
annotations/giantsteps/ annotations in the GiantSteps project format
annotations/.git/       git repo with annotations 
splits/                 file splitting definitions
Sebastian Böck's avatar
Sebastian Böck committed
21
```
Sebastian Böck's avatar
Sebastian Böck committed
22
23
24
25

To convert the audio files to .wav use 'flac -d *.flac'

If you use this dataset for comparison of the results, please cite the above
Sebastian Böck's avatar
Sebastian Böck committed
26
paper. The same dataset is also used in [2]_ and various other papers.
Sebastian Böck's avatar
Sebastian Böck committed
27

Sebastian Böck's avatar
Sebastian Böck committed
28
- ah_* taken from the data set introduced by Andre Holzapfel in:
Sebastian Böck's avatar
Sebastian Böck committed
29
30
31
32
  Three dimensions of pitched instrument onset detection.
  A. Holzapfel, Y. Stylianou, A.C. Gedik, and B. Bozkurt.
  IEEE Transactions on Audio, Speech, and Language Processing, 18(6), 2010

Sebastian Böck's avatar
Sebastian Böck committed
33
- al_* taken from the data set introduced by Alexandre Lacoste in:
Sebastian Böck's avatar
Sebastian Böck committed
34
35
36
37
38
39
40
41
  A supervised classification algorithm for note onset detection.
  A. Lacoste, and D. Eck.
  EURASIP Journal on Applied Signal Processing, 1, 2007.
  It can be downloaded from:
  <http://w3.ift.ulaval.ca/~allac88/dataset.tar.gz>.
  The audio files are taken from the ISMIR 2004 ballroom data set located
  at: <http://mtg.upf.edu/ismir2004/contest/tempoContest/node5.html>.

Sebastian Böck's avatar
Sebastian Böck committed
42
- api_*
Sebastian Böck's avatar
Sebastian Böck committed
43
44
45
  files / annotations by Antonio Pertusa Ibáñez
  http://grfia.dlsi.ua.es/cm/worklines/pertusa/onset/ODB/

Sebastian Böck's avatar
Sebastian Böck committed
46
- ff123_*
Sebastian Böck's avatar
Sebastian Böck committed
47
48
49
  samples for testing audio codecs
  retrieved from http://ff123.net/samples.html

Sebastian Böck's avatar
Sebastian Böck committed
50
- gs_*
Sebastian Böck's avatar
Sebastian Böck committed
51
52
  files by Georgios Siamantas

Sebastian Böck's avatar
Sebastian Böck committed
53
- jpb_* taken from the data set introduced by Juan Pablo Bello in:
Sebastian Böck's avatar
Sebastian Böck committed
54
55
56
57
  A tutorial on onset detection in music signals.
  J. Bello, L. Daudet, S. Abdallah, C. Duxbury, M. Davies, and M. Sandler.
  IEEE Transactions on Speech and Audio Processing, 13(5), 2005

Sebastian Böck's avatar
Sebastian Böck committed
58
- lame_*
Sebastian Böck's avatar
Sebastian Böck committed
59
60
61
  files used by the lame dev crew to test the lame mp3 encoder
  retrieved from http://lame.sourceforge.net/quality.php

Sebastian Böck's avatar
Sebastian Böck committed
62
- mck_*
Sebastian Böck's avatar
Sebastian Böck committed
63
64
65
66
  public part of Martin McKinney's beat/tempo set
  retrieved from:
  http://www.music-ir.org/mirex/wiki/2006:Audio_Tempo_Extraction

Sebastian Böck's avatar
Sebastian Böck committed
67
- mit_*
Sebastian Böck's avatar
Sebastian Böck committed
68
69
70
  files taken from the MIT media lab
  retrieved from http://sound.media.mit.edu/media.php

Sebastian Böck's avatar
Sebastian Böck committed
71
- sb_* taken from the data set introduced by Sebastian Böck in:
Sebastian Böck's avatar
Sebastian Böck committed
72
73
74
75
76
77
  Universal Onset Detection with Bidirectional Long Short-Term Memory Neural
  Networks.
  F. Eyben, S. Böck, B. Schuller, and A. Graves.
  Proceedings of the 11th International Society for Music Information Retrieval
  Conference (ISMIR), 2010.

Sebastian Böck's avatar
Sebastian Böck committed
78
- SoundCheck2_*
Sebastian Böck's avatar
Sebastian Böck committed
79
80
  files taken from the Alan Parson Sound Check 2 audio test CD

Sebastian Böck's avatar
Sebastian Böck committed
81
- vorbis_*
Sebastian Böck's avatar
Sebastian Böck committed
82
83
84
  files used for a OGG Vorbis listening test
  retrieved from http://hem.passagen.se/ingets1/vorbis.htm

Sebastian Böck's avatar
Sebastian Böck committed
85
86
87
88
- The origin of the violin sample is unknown.

References
==========
Sebastian Böck's avatar
Sebastian Böck committed
89

Sebastian Böck's avatar
Sebastian Böck committed
90
91
92
93
[1]: Sebastian Böck, Florian Krebs and Markus Schedl,
    *Evaluating the Online Capabilities of Onset Detection Methods*,
    Proceedings of the 13th International Society for Music Information
    Retrieval Conference (ISMIR), 2012.
Sebastian Böck's avatar
Sebastian Böck committed
94

Sebastian Böck's avatar
Sebastian Böck committed
95
96
97
98
[2]: Sebastian Böck and Gerhard Widmer,
    *Maximum Filter Vibrato Suppression for Onset Detection*,
    Proceedings of the 16th International Conference on Digital Audio Effects
    (DAFx), 2013.