Handling Time-Derivative Features in a Missing Data Framework for Robust Automatic Speech Recognition | IEEE Conference Publication | IEEE Xplore