Human Behaviour in HCI: Complex Emotion Detection through Sparse Speech Features

摘要: To obtain a more human-like interaction with technical systems, those have to be adaptable the users' individual skills, preferences, and current emotional state. In human-human (HHI) behaviour of speaker is characterised by semantic prosodic cues, given as short feedback signals. These signals minimally communicate certain dialogue functions such attention, understanding, confirmation, or other attitudinal reactions. Thus, these play an important role in progress coordination interaction. They allow partners inform each their behavioural affective state without interrupting ongoing dialogue. Vocal communication provides acoustic details revealing speaker's feelings, believes, social relations. Incorporating discourse particles (DPs) human-computer (HCI) systems will detection complex emotions, which are currently hard access. Complex emotions turn closely related human behaviour. Hence, integrating automatic DP emotion assignment HCI first approach integration understanding systems. In this paper we present methods allowing extract pitch-contour DPs assign observed DPs. We investigate occurrences naturalistic show that may assigned automatically. Furthermore, indeed behaviour, showing age-gender specific usage during HCI. Additionally, prove used automatically detect classify