Auditory Attention Detection Dataset Nanjing University

Auditory Attention Detection Dataset

The experiment is conducted by Yuanming, Zhang and Ziyan, Yuan, supervised by Jing, Lu.

Two zip files are included in this dataset.

The script.zip file contains several scripts to divide each trial into several parts. These scripts are useful if you want to split the dataset into training, validation and test set.

The other zip file contains the preprocessed EEG and audio signals. (Not the raw EEG!)

Each mat file contains EEG and audio signals from one subject. Their are two variables named expinfo and data.

The expinfo contains the attended direction information and audio stimulus playback information of each trial of the experiments. The azimuth column in expinfo indicates the directions of the two competing speakers. The negative directions indicated the speaker coming from the left, and positive coming from the right.

The data contains several essential fields relative to the EEG and audio signals. The data.eeg field is a cell array with 32 cell elements. Each cell contains the EEG signals in a single trial. And the index of the cell is the number of trial. The data.leftWav and data.rightWav field are original audio stimuli, and the data.leftEnv and data.rightEnv are the envelopes of the data.leftWav and data.rightWav respectively.

The data.dim field tells you the physical meaning of each dimension of each signal, and the data.dim.chan.eeg contains the name of each channel of the EEG signal.