4562892
doi
10.5281/zenodo.4562892
oai:zenodo.org:4562892
user-dcase
Salamon, Justin
Adobe Research, San Francisco CA, United States
Shah, Ankit
Language Technologies Institute, Carnegie Mellon University, Pittsburgh PA, United States
Wisdom, Scott
Google, Inc
Hershey, John
Google, Inc
Erdogan, Hakan
Google, Inc
Serizel, Romain
Université de Lorraine, CNRS, Inria, Loria, F-54000 Nancy, France
DESED_synthetic
Turpault, Nicolas
Université de Lorraine, CNRS, Inria, Loria, F-54000 Nancy, France
url:https://hal.inria.fr/hal-02160855v2
url:https://hal.inria.fr/hal-02355573
info:eu-repo/semantics/openAccess
Creative Commons Attribution 4.0 International
https://creativecommons.org/licenses/by/4.0/legalcode
DCASE
Sound event detection
<p>Link to the associated github repository: <a href="https://github.com/turpaultn/Desed">https://github.com/turpaultn/Desed</a></p>
<p>Link to the papers: <a href="https://hal.inria.fr/hal-02160855"><em>https://hal.inria.fr/hal-02160855</em></a>, <a href="https://hal.inria.fr/hal-02355573v1">https://hal.inria.fr/hal-02355573v1</a></p>
<p>Domestic Environment Sound Event Detection (DESED).</p>
<p><strong>Description</strong><br>
This dataset is the synthetic part of the DESED dataset. It allows creating mixtures of isolated sounds and backgrounds.</p>
<p>There is the material to:</p>
<ul>
<li>Reproduce the DCASE 2019 task 4 synthetic dataset</li>
<li>Reproduce the DCASE 2020 task 4 synthetic train dataset</li>
<li>Creating new mixtures from isolated foreground sounds and background sounds.</li>
</ul>
<p><strong>Files:</strong></p>
<p><strong>If you want to generate new audio mixtures yourself from the original files.</strong></p>
<ol>
<li><strong>DESED_synth_soundbank.tar.gz</strong> : Raw data used to generate mixtures.</li>
<li><strong>DESED_synth_dcase2019jams.tar.gz</strong>: JAMS files, metadata describing how to recreate the dcase2019 synthetic dataset<strong> </strong></li>
<li><strong>DESED_synth_dcase20_train_val_jams.tar: </strong>JAMS files, metadata describing how to recreate the dcase2020 synthetic train and valid dataset.</li>
<li><strong>DESED_synth_dcase20_eval_jams.tar: </strong>JAMS files, metadata describing how to recreate the dcase2020 synthetic eval dataset (only the basic one, variants of it have been made but not presented here).</li>
</ol>
<p><strong>If you simply want the evaluation synthetic dataset used in DCASE 2019 task 4.</strong></p>
<ol>
<li><strong>DESED_synth_eval_dcase2019.tar.gz</strong><strong> </strong>:<strong> </strong>Evaluation audio and metadata files used in dcase 2019 task 4.</li>
</ol>
<p> </p>
<p>The mixtures are generated using Scaper (https://github.com/justinsalamon/scaper) [1].</p>
<p>* Background files are extracted from SINS [2], MUSAN [3] or Youtube and have been selected because they contain a very low amount of our sound event classes.<br>
* Foreground files are extracted from Freesound [4][5] and manually verified to check the quality and segmented to remove silences.</p>
<p><strong>References</strong><br>
[1] J. Salamon, D. MacConnell, M. Cartwright, P. Li, and J. P. Bello. Scaper: A library for soundscape synthesis and augmentation<br>
In IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), New Paltz, NY, USA, Oct. 2017.</p>
<p>[2] Gert Dekkers, Steven Lauwereins, Bart Thoen, Mulu Weldegebreal Adhana, Henk Brouckxon, Toon van Waterschoot, Bart Vanrumste, Marian Verhelst, and Peter Karsmakers.<br>
The SINS database for detection of daily activities in a home environment using an acoustic sensor network.<br>
In Proceedings of the Detection and Classification of Acoustic Scenes and Events 2017 Workshop (DCASE2017), 32–36. November 2017.</p>
<p>[3] David Snyder and Guoguo Chen and Daniel Povey.<br>
MUSAN: A Music, Speech, and Noise Corpus.<br>
arXiv, 1510.08484, 2015.</p>
<p>[4] F. Font, G. Roma & X. Serra. Freesound technical demo. In Proceedings of the 21st ACM international conference on Multimedia. ACM, 2013.<br>
<br>
[5] E. Fonseca, J. Pons, X. Favory, F. Font, D. Bogdanov, A. Ferraro, S. Oramas, A. Porter & X. Serra. Freesound Datasets: A Platform for the Creation of Open Audio Datasets.<br>
In Proceedings of the 18th International Society for Music Information Retrieval Conference, Suzhou, China, 2017.</p>
<p> </p>
Zenodo
2020-03-07
info:eu-repo/semantics/other
3550598
user-dcase
v2.3
1644433732.487417
7710291574
md5:e1aad0a714bb98d2b58f3d62122077b8
https://zenodo.org/records/4562892/files/DESED_synth_eval_dcase2019.tar.gz
25859
md5:2eba5a6fe230baecc1803dab526a77a5
https://zenodo.org/records/4562892/files/soundbank_validation.tsv
2422047310
md5:03b51e3506ae28157a26101748045e90
https://zenodo.org/records/4562892/files/DESED_synth_soundbank.tar.gz
325956
md5:105774e4528b266c829f3a6fdad4397d
https://zenodo.org/records/4562892/files/DESED_synth_dcase20_eval_jams.tar.gz
1154751
md5:01f2ba4e33c82006d8e407b75f103fe7
https://zenodo.org/records/4562892/files/DESED_synth_dcase20_train_val_jams.tar.gz
18874495973
md5:99cbb7b21299cd473e4acedfd5ad614f
https://zenodo.org/records/4562892/files/dcase21_synth.tar.gz
3096604
md5:e5d6348d9b9ca19d08b7afba0e987de3
https://zenodo.org/records/4562892/files/DESED_synth_dcase2019jams.tar.gz
public
https://hal.inria.fr/hal-02160855v2
Is supplement to
url
https://hal.inria.fr/hal-02355573
Is supplement to
url
10.5281/zenodo.3550598
isVersionOf
doi