Using Gaussian Mixture Model And Partial Least Squares Regression Classifiers For Robust Speaker Verification With Various Enhancement Methods