Answers are grounded in this paper and cite their sources.
Voice assistants are increasingly integrated into users’ daily lives, yet little is known about how their synthetic voices convey emotions and influence perceptions of humanity and trust. This study addresses that gap by comparing the emotional expressiveness of Siri, Alexa and Google Assistant and advancing an emotionally grounded Computational Trust Model based on ERT. The model conceptualises trust as an affective outcome in which vocal expressiveness triggers emotional responses, measured by valence and arousal, that shape perceived human-likeness and trust. An experimental study with 900 participants randomly exposed to one of the three voices delivering positive or negative product reviews examined these processes. After exposure, participants rated their emotional reactions, the assistant’s human-likeness, and trust. Results showed that expressive, positively valenced voices, especially Siri’s, evoked stronger emotional responses, higher perceived humanity and greater trust. Automatic Emotion Detection confirmed Siri’s superior emotional expressiveness. Positive messages enhanced emotional engagement, whereas negative valence had weaker effects. By extending ERT to voice-based interactions, the study introduces a scalable trust model rooted in emotion. It advances Human–Computer Interaction research by revealing how synthetic voices can express basic emotions and trigger affective responses, offering practical applications in customer service, education and healthcare.
No takes yet. Share an insight, caveat, or question.
Calahorra-Candao et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: