Determining prosodic markings for text-to-speech systems - using neural network to determine prosodic markings based on linguistic categories such as number, verb, verb particle, pronoun, preposition etc.

摘要

The method involves using a neural network (1) to determine prosodic markings e.g. phrase boundaries based on linguistic categories. The markings are evaluated. The linguistic categories of at least three words of a text to be synthesised are applied to the input of the neural network. The characteristics of each prosodic marking are detected by neuronal auto-associators (7) which are trained respectively for a certain prosodic marking. The output information from each auto-associator is evaluated in a neural classifier.