Related Experiment Video
Updated: Jan 17, 2026

Creating Virtual-hand and Virtual-face Illusions to Investigate Self-representation
Published on: March 1, 2017
Voice clones sound realistic but not (yet) hyperrealistic
Nadine Lavan1, Mairi Irvine1, Victor Rosi2
1Centre for Brain and Behaviour, School of Biological and Behavioural Sciences, Queen Mary University of London, London, United Kingdom.
AI-generated voices can sound as real as human voices, making them hard to distinguish. However, a "hyperrealism effect" seen in AI faces was not observed in AI voices, which were perceived as more dominant.
Area of Science:
- Human-Computer Interaction
- Speech Processing
- Psychology
Background:
- AI-generated voices are becoming common in daily life.
- Previous research found a "hyperrealism effect" in AI-generated faces.
- The perception of AI-generated voices requires further investigation.
Purpose of the Study:
- To determine if AI-generated voices exhibit a "hyperrealism effect."
- To assess human listeners' ability to distinguish AI voices from human voices.
- To examine perceived social traits (trustworthiness, dominance) of AI and human voices.
Main Methods:
- Tested AI voices generated with and without human counterparts (voice clones vs. latent space models).
- Evaluated listener's ability to discern AI from human voices.
- Assessed perceived trustworthiness and dominance of voices.
Main Results:
- Voice clones were as realistic as human voices, making them difficult to distinguish.
- No "hyperrealism effect" was observed for AI-generated voices.
- AI voices were perceived as more dominant than human voices; some were also seen as more trustworthy.
Conclusions:
- Current AI voice technology does not produce a "hyperrealism effect" comparable to AI faces.
- Differences in perception between voice and image may explain the lack of hyperrealism.
- Realistic AI voices present risks for misinformation and fraud, but also offer beneficial applications.
Related Concept Videos
Modeling and Similitude
Nonconscious Mimicry
Unrealistic Optimism Bias
Facial Feedback Hypothesis
Impression Management Techniques IV: Altercasting
Auditory Perception

