Speech masker formalism : a rule formalism operating on a multi-level, synchronized data structure

In this paper the Speech Maker Formalism (SMF) is described. Speech Maker is a general framework for text-to-speech synthesis (van Leeuwen & te Lindert, 1993. Computer Speech and Language 7, 149-167), and SMF is operational in this context. SMF has been developed as a general-purpose rule formalism for the application domain of text-to-speech synthesis. A large range of symbolic transformations that a linguist may want to specify in this domain can be formulated in SMF. A central characteristic of SMF is that it is two-dimensional and that the rules closely resemble the two-dimensional data structure they operate on. An extensive and detailed description is given of the possibilities of SMF, concerning both the specification of patterns in the data structure and the alteration of its contents. In particular, the completeness and the expressive power of the formalism are discussed.