Virtual Story Telling for Kids: Expressive and Cognitive Aspects of Voice Synthesis – EXOVOICES

Overexposure to screens has harmful consequences for children's cognition, while audio stories can have a positive impact on their attentional and imaginative engagement. Virtual reading of stories by synthetic voices can improve immersion and imagination by e.g. allowing children to be addressed by their first name. As current voice synthesizers suffer from monotony and lack of expressiveness, EXOVOICES proposes a neural synthesizer augmented by automatic linguistic annotation of text (structural, semantic and expressive content). The generated synthetic voice will be more diversified, according to the structure of the story and to the expressive content carried by the text. Current evaluation protocols are not adapted to emphasize the monotony or expressivity of a synthetic voice. EXOVOICES proposes to contribute to the definition of expressivity and will develop an experimental protocol that will allow the evaluation of this notion on natural and synthetic voices. The impact of listening to natural and synthetic stories on children's attention and imagination will be assessed through cognitive experiments, which will constitute a new paradigm for the evaluation of voice synthesis. A better understanding of attention fluctuation mechanisms during reading and listening will have important implications in education and for the care of children who suffer from attention deficit. Verbal human / machine communication enhancement allowed by EXOVOICES will have a strong impact on futur vocal interfaces development. New kinds of interactivity in recreational or educational audio applications will be possible, which will help for screen time reduction purposes.