High-Dimensional Linear Representations for Robust Speech Recognition

doi 10.1109/ita.2010.5454172