I found in compute-meanvar-stats.py it uses standard scaler to do normalization. For spectrogram, I am not sure if it is right to do normalization on 2D matrix. I have observed that if we apply normalization on 2D matrix, it is a little weird. However, if applied to 1D vector (ravel 2D matrix to 1D) it makes more sense. Do you consider this situation?
I found in compute-meanvar-stats.py it uses standard scaler to do normalization. For spectrogram, I am not sure if it is right to do normalization on 2D matrix. I have observed that if we apply normalization on 2D matrix, it is a little weird. However, if applied to 1D vector (ravel 2D matrix to 1D) it makes more sense. Do you consider this situation?