Irene Martin Morato
Manu Harju
Annamaria Mesaros
2021-08-27
<p>The dataset was created for studying estimation of strong labels using crowdsourcing.</p>
<p>It contains 20 synthetic audio files created using Scaper, the reference annotation created with Scaper, and the annotation outcome. Annotation was performed using Amazon Mechanical Turk.</p>
<p>Audio files contain excerpts of recordings uploaded to freesound.org.(from Urban Sound 8k dataset). Please see FREESOUNDCREDITS.txt for an attribution list. </p>
<p>The dataset contains: </p>
<ul>
<li>audio: the 20 synthetic soundscapes, each 3 min long</li>
<li>ground truth: the "true" reference annotation created using Scaper</li>
<li>estimated strong labels: the reference annotation created from the crowdsourced data</li>
<li>audio tags: the weak labels corresponding to each 10 s segment of the soundscapes, as annotated</li>
</ul>
<p>For details on the annotation procedure and label processing methodology, see the following paper:</p>
<p>Irene Martin Morato, Manu Harju, and Annamaria Mesaros. <em><a href="https://arxiv.org/abs/2107.12089">Crowdsourcing strong labels for sound event detection</a>.</em> In IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA 2021). New Paltz, NY, Oct 2021.</p>
<p> </p>
<p> </p>
https://doi.org/10.5281/zenodo.5126478
oai:zenodo.org:5126478
Zenodo
https://zenodo.org/communities/machine-listening-tau
https://doi.org/10.5281/zenodo.5126477
info:eu-repo/semantics/openAccess
Other (Non-Commercial)
MAESTRO Synthetic - Multi-Annotator Estimated Strong Labels
info:eu-repo/semantics/other