Преобразование букв в фонемы

We present a discriminative structure- prediction model for the letter-to-phoneme task, a crucial step in text-to-speech process- ing. Our method encompasses three tasks that have been previously handled separately: input …...