Spraakmaker: a text-to-speech system for the Dutch language

Spraakmaker integrates a number of modules resulting from the combined efforts of several institutes in the Netherlands which aim at doing research in the field of automatic text-to-speech conversion. Spraakmaker is intended as a demonstration model for Dutch as well as a flexible tool for future research. In its design, Spraakmaker differs from most other existing text-to-speech systems. This paper first discusses the general language-independent framework on which Spraakmaker is built, which concerns the data structure used to store relevant data for the text-to-speech process and the general means of control. Then Spraakmaker is discussed as a specific implementation, which illustrates the use of the data structure and how linguistic modules operate on this structure.