Dataset Open Access
TUT Acoustic Scenes features dataset consists of feature matrices extracted for 10-seconds audio segments from 15 acoustic scenes:
Dataset is split into **Train set** and **Test set**. Each acoustic scene in the Train set has 300 segments, and Test set has 100 segments. The dataset contains similar material to other TUT Acoustic Scenes 2017 datasets: TUT Acoustic Scenes 2017 development dataset and TUT Acoustic Scenes 2017 evaluation dataset. All these datasets are composed from same pool of original audio recordings, but exact audio segments selected for the datasets differ.