This paper describes the collection and analysis of a multi-style emotional speech corpus, accomplished to study the variations of some acoustical parameters. Specifically, three emotional styles were considered: happiness, sadness and anger. Speech data in a neutral style were also collected, and prosodic differences of each style with respect to this neutral baseline were quantified. According to the analysis results, experiments were also carried out to synthesize emotional styles by modifying prosodically neutral signals both extracted from the original corpus and produced by our Text To Speech synthesis system. Perceptual tests were made to evaluate the effectiveness of the adopted method.