Faculty of Applied Sciences Communication of Emotions for E-creatures

from Anty Foundation and all the other researchers in VUB ETRO-DSSP for their valuable contribution and support throughout this project. Abstract The goal of this project is to produce affective speech that can convey natural vocal emotions in a non-existing and non-understandable language. The system uses a database with natural expressive speech samples from a professional speaker and a database with naturally spoken neutral speech samples from the same speaker. To produce nonsense utterance with a given desired emotion, the synthesizer randomly selects an expressive speech sample of the proper type from the expressive database and uses this as a prosodic template. Next, a nonsense carrier phrase is constructed that has the same segmental structure as the selected prosodic template. This is done by concatenating segments from the neutral database. Then, the same pitch and timing structure as found in the prosodic template is copied on the nonsense carrier phrase through prosody transplantation process which uses PSOLA (Pitch Synchronous Overlap-Add) method. Finally, voice quality structure of the template emotional utterance is applied to the artificially generated speech, which results as the final synthesized emotional speech. 4 Declaration Here by I declare that this thesis was composed by me, that the work contained herein is my own except where explicitly stated otherwise in the text, and that this work has not been submitted for any other degree or professional qualification except as specified.