A German text-to-speech system, called FLEX-DEUTSCH, which has been developed at the Institute of Linguistics (Phonetic Laboratory) of the Hungarian Academy of Sciences, Budapest, was used to generate CV- and VCV-stimuli. These stimuli were administered to 20 German speaking subjects in order to evaluate the system's efficiency concerning consonantal segments. The results of this perception test will be the basis for further modification of the acoustic parameters to improve the segmental intelligibility of the system.