Gibberish speech as a tool for the study of affective expressiveness for robotic agents

Recent technological advancements bring virtual agents, avatars, and social robotic characters into our daily lives. These characters must acquire the ability to express (simulated) emotions vocally and gesturally. In the vocal channel, Natural Language Interaction technologies today have some limitations when used in real-world natural environments and the models of expressivity of text to speech synthesis engines are not yet mature enough. To address these limitations, an alternative form of vocal communication - gibberish speech - is introduced in this paper. Gibberish speech consists of vocalizations of meaningless strings of speech sounds, and thus has no semantic meaning. It is occasionally used by performing artists or for cartoon animations and games to express intended emotions (e.g. Teletubbies and The Sims). In this paper, our approach for constructing expressive gibberish speech is described and the experimental evaluations with its intended robotic agents are reported. It is shown that the generated gibberish speech can contribute to a significant extent to studies concerning emotion expression for robotic agents and can be further utilized in affective human-robot interaction studies