Robot Voices Matter More Than You Think: From Gendering to Inclusion with Gender-Ambiguous Voices

Licentiate thesis, 2026

Voice shapes social perception quickly and persistently. Beyond transmitting content, it shapes impressions of gender, age, origin, warmth, competence, and authority, which in turn affect how speakers and technologies are judged. In Human-Robot Interaction (HRI) and Human-Computer Interaction (HCI), robots and voice assistants commonly use clearly masculine or feminine synthetic voices. This practice reflects a broader tendency to rely on binary gender cues, both as a general design convention and as a means of making technology appear natural and acceptable. However, these design defaults can reinforce gender stereotypes and role expectations, contribute to masculine-by-default perceptions of technology, and make interactions less welcoming for users who do not align with binary categories. This thesis examines gender-ambiguous synthetic voices as a design intervention that resists clear categorisation within masculine or feminine norms, investigating how such voices are perceived, whether they support inclusion, and whether they influence robot gendering.

This thesis comprises four papers. Paper I presents a systematic review of voices beyond the binary in HCI, revealing inconsistent terminology, limited transparency about voice availability, and a lack of methodological consistency in evaluation. Paper II reports a large-scale perception study evaluating ambiguous voices across trustworthiness, comfort, appeal, anthropomorphism, and aversion, showing that these voices are not experienced uniformly and can elicit more critical responses from nonbinary listeners. Paper III and Paper IV examine how voice influences imagined and constructed robot form using sketching and physical prototyping as methods. Together, these studies show that gender-ambiguous voices can reduce explicit gender attribution and encourage less clearly gendered embodiments, while a masculine-by-default tendency remains.

Overall, this thesis shows that gender-ambiguous voices can soften binary gendering in interaction and robot design, but they do not provide identity-based inclusion without community-informed approaches and careful framing.