Feature-based pronunciation modeling with trainable asynchrony probabilities (2004)

by K Livescu, J Glass
Venue:in Proc. International Conference on Spoken Language Processing (ICSLP