University of Twente Student Theses


Exploring Roboticness and Applicability of Voices to Social Robots

Labedzka, P.Z. (2024) Exploring Roboticness and Applicability of Voices to Social Robots.

[img] PDF
Abstract:Robots, especially social robots, are becoming increasingly popular in today’s world. With that, the need to understand the perception of robotic speech and its appropriate design arises. This project aims to understand and model what makes a voice perceived as ‘robotic’ as well as what makes it suitable for social robots. This research has been split into two studies. In the first one, people’s impressions of roboticess and the suitability of voices to social robots, along with associations with robots, have been gathered in a form of an online survey. In the second one, an attempt was made to quantify and measure those impressions using speaker embeddings, fixed-dimensional vectors that capture the speaker’s identity, and vocal parameters. The results support the earlier findings that the less robotic a voice is, the more applicable it becomes to a social robot. Moreover, Study I also revealed that people still often associate robots with those from the media. This suggests that media is a useful source of knowledge for understanding what makes a voice sound robotic. The findings of Study II further support this conclusion, as it was found that vocal parameters related to fundamental frequency and loudness are the ones primarily related to the roboticness of a given voice. On the other hand, thematic analysis of perceived factors contributing to roboticness, gathered in Study I, revealed less measurable descriptors of voices with such quality. Henceforth, in Study II, speaker embeddings were employed to explore whether they encode roboticness. The results of this study indicated that speaker embeddings contained information related to the perception of robotic voices to some extent. However, this finding needs validation through further research, which should address significant limitations not covered in this study.
Item Type:Essay (Bachelor)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Subject:02 science and culture in general, 05 communication studies, 54 computer science
Programme:Technology and Liberal Arts & Sciences BSc (50427)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page