A particular problem for the automatic prediction of prosody in speech synthesis is the realisation of accented syllables since these are affected by many parameters and are perceptually very salient. For the Portuguese language, in Europe, a set of comprehensive quantitative characterisation data and rules is totally lacking. The present paper is intended to be a quantitative contribution to the solution of this problem. In this paper, a preliminary modelling of duration, intensity and variation of F0 in the tonic syllable will be presented. The dependencies of the model with the syllable position in the word and the word position in the phrase are also presented.
