Robustness-Related Issues in Speaker Recognition