PUB - Publikationen an der Universität Bielefeld

This article presents a phonetic and perceptive transcription system for German speech melody. Following the IPO-approach for the description of intonation, the phonetically defined categories of the melodic labelling procedure are based on a model which includes all perceptively relevant German pitch movements for read speech. To allow the application of the model-based categories as units of an annotation system for read and spontaneous German speech melody, two requirements were fulfilled: one is the experimental evaluation of the validity of the model for spontaneous German utterances and the other is the development of an appropriate labelling and testing environment. This labelling and testing environment offers an auditive and visual control of the melodic annotation. By an implementation of the model as part of the environment and by using the control facilities of the labelling environment the melodic description in question is intersubjective more reliable.