An overview of the current state of the SVOX TTS system for German is presented in this article. A detailed presentation of the text analysis component is given and it is shown how word and sentence structure analysis and even number-to-phoneme conversion and grapheme-to-phoneme conversion can be expressed concisely and in an easily readable way in a definite-clause-grammar (DCG) formalism, which is interpreted by a chart-parser. We also give an overview of the prosody control in the system, which consists of rule-based accentuation and prosodic phrasing and of a purely statistical and neural-net-work-based acoustic interpretation of suprasegmental phono-logical information.

Keywords: text-to-speech synthesis, prosody control, syntactic analysis, grapheme-to-phoneme conversion