N06-4011 Speaker Adaptation In the context of speaker-adaptive training , we use two forms of feature-space
N04-4007 Length Normalization ( VTLN ) , Speaker-Adaptive Training ( SAT ) and iterative unsupervised
hide detail