The SYRUB text-to-speech system had been designed to translate unrestricted German text into a sequence of parameters that can be used to drive different speech synthesizers (ref 1). The interface parameters consist of the phoneme code, fundamental frequency (f0), sound duration, and sound intensity. For synthesizers that do not operate with phonemes as the basic units additional information, e.g., for controlling coarticulation, is available. To produce a fairly natural intonation, several steps are required: a morphemic analysis generates information for phonemization, word stress assignment, and segmentation into phonetic syllables. An end grapheme analysis supplies word classes needed to mark phrase boundaries. For f0 assignment a declination line and stress patterns are applied. Sound duration is governed either by context-dependent rules or by isochrony, generating rhythmic speech.