Analyzing the Use of Auditory Filter Models for Making Interpretable Convolutional Neural Networks for Speaker Identification.