This code implements sound event detection in audio files using spatial features. The features contain generalized cross correlation, auto correlation, and energy of the signal in different subbands. The model is a convolutional recurrent nueral network followed by some LSTM modules.
File 'data4mynet2.mat' contains all training data, test data and the labels. I can share the data upon request