On the use of stochastic inference networks for representing multiple word pronunciations.

Renato De Mori, Charles Snow, Michael Galler

Browse the full ICASSP paper archive.