DCASE 2013 stands for Detection and Classification of Acoustic Scenes and Events, which is a dataset for sound event detection. The dataset consists of pure audio recordings, and its characteristic is that a single sound event stands out in an acoustic scene.

In the dataset for the scene classification task, each sound scene consists of a 30-second recording. The dataset will consist of 2 parts, each consisting of 6 audio recordings for each scene (class). The list of scenes includes 10 categories: busy street, quiet street, park, open-air market, bus, subway train, restaurant, store/supermarket, office, subway station.