Combining Discriminative Feature, Transform, and Model Training for Large Vocabulary Speech Recognition | IEEE Conference Publication | IEEE Xplore