This paper examines how voice labels such as “ambiguous”, “queer”, and “nonbinary” shape the design and interpretation of synthetic voices in HCI. Drawing on feminist and queer theory, the work argues that voice is not neutral but socially and politically constructed, and that the terms used to describe beyond-binary voices carry ethical and representational consequences. To ground the theoretical discussion, the study includes a survey with nonbinary participants to explore which labels feel most affirming for beyond-binary voices.

Objectives

1. Examine how feminist and queer theory can inform the labelling of synthetic voices in HCI.

2. Analyse the conceptual differences between “ambiguous”, “queer”, and “nonbinary” as voice design labels.

3. Investigate how nonbinary individuals label gender-expansive synthetic voices.

Results

The paper showed that voice labels such as “ambiguous”, “queer”, and “nonbinary” are often treated as interchangeable in HCI, despite serving different purposes. Survey responses from nonbinary participants indicated a stronger use of gender-expansive terms like “nonbinary” and “any gender”, with few relying solely on binary labels. These findings highlight the need for clearer and more community-informed terminology to support ethical and inclusive voice design.

This study investigates how robot voices shape physical embodiment by reversing the traditional design process: instead of assigning a voice to an already designed robot, participants first listened to a voice and then built a physical prototype. Using the RoMix toolkit, this research examines how ambiguous, feminine, and masculine voices influence anthropomorphism and gendered design choices in tangible robot forms.

Objectives

1. Examine whether voice type affects the level of anthropomorphism (human-like vs. machine-like features) in constructed robots.

2. Explore whether voice type influences the inclusion of gendered characteristics in robot design.

3. Compare explicit self-reports (ratings) with implicit design outcomes (selected physical components).

Results

Voice type (ambiguous, masculine, feminine) influenced how participants physically built robots, even though explicit ratings showed no clear differences. While participants described their robots as neutral, their design choices revealed implicit patterns shaped by the voice condition. Ambiguous voices led to fewer human-like and more hybrid forms, whereas masculine voices encouraged more human-like designs. Despite reported neutrality, the ambiguous condition resulted in more masculine torso choices, highlighting a persistent masculine-by-default tendency that was not reflected in self-reports.

This paper investigates how people of different gender identities (nonbinary, men, women) perceive artificial gender-ambiguous voices (voices blending masculine and feminine traits) across dimensions beyond gender. By comparing six ambiguous voices and testing priming effects, it explores inclusivity in human-computer interaction.

Objectives

1. Compare perceptions of ambiguous voices across trust, appeal, comfort, anthropomorphism, and aversion.

2. Test if telling participants about a voice’s ambiguity changes evaluations.

3. Inform inclusive, bias-resistant voice design.

Results

All six voices were rated differently. The British-accented voice (“Fable”) was most trusted and liked. Non-binary participants rated voices less trustworthy and more aversive than men and women, suggesting current ambiguous designs do not yet resonate. Priming had little positive effect and sometimes increased criticism. Findings show that ambiguity alone does not ensure inclusivity: accent, cultural fit, and community-informed design also matter.

TO THE PAPER

Voice-enabled devices like virtual assistants and robots often use distinctly masculine or feminine voices, reinforcing gender stereotypes and excluding non-binary identities. Gender-ambiguous voices, which blend masculine and feminine traits, offer a more inclusive alternative. This study conducts the first systematic review of gender-ambiguous voices in Human-Computer Interaction (HCI), analyzing 36 studies to explore their definitions, creation methods, and user perceptions. It highlights research gaps and provides recommendations for designing inclusive voice technologies and advancing equitable human-technology interactions. 

Objectives

1. Define and standardise the concept of gender-ambiguous voices in the HCI literature.

2. Assess the availability, characteristics, and creation methods of these voices.

3. Explore how users perceive these voices and the methods used to evaluate them.

4. Provide actionable recommendations for designing and adopting inclusive voice technologies.

The review revealed inconsistent definitions and limited standardisation of gender-ambiguous voices in HCI. Most studies relied on modifying existing technologies rather than developing new voices, with few offering open-source solutions. User perceptions varied widely, with some voices seen as inclusive but others criticised as artificial or less desirable. The findings emphasise the need for clear guidelines, better creation methods, and standardised evaluation metrics to support the adoption of gender-ambiguous voices and promote inclusivity in voice-enabled technologies.

TO THE PAPER

Unlike previous studies that focus on gendered design elements, this work uniquely investigates the impact of gender-ambiguous voices, blending masculine and feminine traits, on how people imagine robots in gendered occupational contexts. By analysing sketches and responses, this study contributes new insights into reducing gender biases and promoting inclusive robot design.

Objectives

1. Investigate whether gender-ambiguous voices reduce gender associations to robots.

2. Examine how auditory cues, such as voice characteristics, influence perceptions of robots in various occupational contexts.

3. Provide insights for designing inclusive robots that minimise gender biases in Human-Robot Interaction.

Results

The study revealed that gender-ambiguous voices reduce gender attribution in robot perception, encouraging more neutral or mixed-gender sketches compared to male or female voices. This, in our opinion, supports using ambiguous voices to challenge stereotypes and foster inclusive human-robot interactions. However, a "male-by-default" bias persisted, with participants often assigning male characteristics even in neutral scenarios. These findings highlight the need for intentional design to minimise gender biases in robotics.

TO THE PAPER

This study investigates the association between a robot’s voice and appearance in Human-Robot Interaction (HRI). Using a mixed-methods approach, it examines how factors such as voice pitch, gender, and robot design influence users' mental images and preferences. Conducted in Sweden and Italy, the study highlights cultural differences in how users match robot voices to appearances, offering insights into designing more intuitive and user-centred robots.

Objectives

1. Explore the relationship between a robot’s voice and its physical appearance.

2. Investigate how human-related factors (gender and nationality) shape perceptions of robots.

3. Examine the influence of voice-related factors, such as pitch and gender, on user preferences.

4. Assess whether the order of stimuli presentation (voice-first vs. image-first) affects voice-appearance associations.

Results

The study revealed that users consistently formed mental images of robots based on voice and appearance cues, with nationality significantly influencing preferences. Swedish participants favored genderless voices, while Italians preferred gendered voices aligning with societal roles. Voice pitch also influenced perceptions, with higher-pitched voices linked to smaller robots and lower-pitched voices to larger ones. These findings underscore the need for culturally sensitive and customizable robot designs to enhance human-robot interactions.

TO THE PAPER