Dataset Open Access
A dataset of synthentic sound scenes created in scaper using real-world recordings. Synthesized scenes are split into train and test such that no original recordings are in both splits.
Dataset created for the task of performing both acoustic scene classification jointly with sound event detection. Time stamped annotations and jams files included.
10 acoustic scene classes, 32 sound event classes.
filename syntax is [sceneLabel_recordingID_polyphonylevel]
Full credits / citation details are below, please contact Yogi at email@example.com with any queries.
Helen L Bear, Ines Nolasco, and Emmanouil Benetos, Towards joint sound scene and polyphonic sound event detection. Interspeech 2019. Graz.