Sound design for HRI: a case study exploring appropriateness, naturalness, and sociality of a zoomorphic robot

Sound design plays a crucial role in Human-Robotic Interaction (HRI). However, manufacturers often overlook the capabilities of the robot to reproduce high-quality audio and use a voice which is inappropriate to the robot and the context in which interaction occurs. In this paper we explore the design of an appropriate robot voice that could soften the anthropomorphisation tendencies of robot users by striking the right balance between human and robotic qualities within the context of specific robotic physical forms. An online survey with 99 participants was carried out to evaluate three different voices for a zoomorphic robot, in three differing listening modalities, to evaluate appropriateness, sociability and naturalness, and in different contexts: educational, care-giving and laboratory. Results showed that a robotic sounding voice performs poorly in all dimensions compared to a more human-like sounding voice. However, by applying sound design techniques to this human-like voice we were able to reduce naturalness (making the voice sound more robotic) while maintaining appropriateness close to the human-like voice. This confirms that it is possible to design robot voices that are less natural and anthropomorphic but still appropriate and foster sociability.