This paper describes the implementation of a text-to-speech system for the Dutch language. Important features by which the system distinguishes itself from most other text-to-speech systems are (a) the central position of a database, (b) the multi-level design of this data structure, and (c) the special form of the rule formalism designed for manipulating the database's contents. The current linguistic analysis modules, and the way they communicate with the database, are described. Examples of the usage of the rule formalism are presented. Finally, the system's interactive usage is briefly touched upon. Keywords: Dutch, grid, rule formalism, Speech Maker, text-to-speech system.