Multi-Dimensional Meaning Annotation in Synthesis of Listener Vocalizations

Abstract

            The meaning annotation of listener vocalizations is a crucial step towards the synthesis of these vocalizations.This chapter describes a systematic study of vocalizations’ meanings. We propose a multi-dimensional annotation approach aimed at obtaining appropriateness ratings of each vocalization for each of the meanings. We conduct a listening test where multiple subjects annotate (characterize) a set of listener vocalizations using a multidimensional set of meaning descriptors. Typical impressions on context-independent meaning of listener vocalizations are being investigated. We also analyse the relevance of behaviour properties for the meaning perception of listener vocalizations.