Deep Neural Network based Text-Dependent Speaker Recognition: Preliminary Results