Speaker and Channel Factors in Text-Dependent Speaker Recognition