Back

Using the obtained Google Faculty Research Award to create FSD, a large-scale audio dataset based on Freesound content

Using the obtained Google Faculty Research Award to create FSD, a large-scale audio dataset based on Freesound content

17.07.2018

The Music Technology Group of the UPF is creating FSD, a large-scale, general-purpose open audio dataset composed of Freesound content annotated with labels from the AudioSet Ontology. The MTG was awarded a Google Faculty Research Award to support this initiative with a proposal developed by Eduardo Fonseca under the supervision of Xavier Serra.

 

With the Google award the MTG is supporting some of the research and development work of the Freesound Datasets platform. To accelerate the creation and improve the quality of FSD, part of this funding is being used to hire summer research assistants to produce high quality annotations as well as doing platform maintenance tasks. There are still few positions available for this summer. Interested candidates can send a mail to [email protected] with a short motivation statement.

 

We believe that FSD will be instrumental to foster research in sound recognition and will have a significant impact in our research community. We launched three months ago the first machine learning Kaggle competition with a small subset of FSD and more than 500 teams are actively participating in it. Check the Freesound General-Purpose Audio Tagging Challenge for more details.

Categories: