https://ieeexplore.ieee.org/document/9756815
The expectations for virtual assistant personalities seem to require a combination of human-like and machine-like characteristics. This results in a wide range of requirements for expressing personality.
In this paper, the authors investigate how verbal and nonverbal communication cues influence the portrayed personality of virtual characters. They examine the impact of factors such as motion and speech realism on personality perception. The study explores how realism affects the perception of virtual characters and agents based on each Big Five personality trait.
To ensure consistency, the researchers controlled for factors that previous studies have shown to influence personality perception, such as clothing and facial features. They also used a diverse range of actors and avatars with different accents and appearances, which sets their work apart from studies that only focus on individual characters of a single sex.
The authors procedurally generated a set of eye-blink intervals that were uniformly applied to each character, following similar approaches in previous work.
The paper provides a brief overview of the study, highlighting the investigation of communication cues and personality portrayal in virtual characters. It also distinguishes itself from previous work by examining the impact of voice and motion on personality in unscripted virtual conversations.
One aspect that stands out is the effort made to maintain consistency in the experiment, considering the subjective nature of personality perception. The researchers also extensively reviewed prior psychology studies to inform the design of their experiment, demonstrating a strong interdisciplinary connection and building upon existing knowledge.
One limitation is that the cultural background of the participants in the experiment is not clear. Different cultures may interpret movements, gestures, and appearances differently, which could impact the results.
To address this, future research could involve participants from different cultures to see how cultural factors influence the perception of robotic motion, TTS voice, and other aspects. This would provide valuable insights into how different cultures perceive these cues and contribute to the field.
In addition, it would be interesting to explore how these techniques for voice and motion affect self-perception of personality when applied to our own voices and avatars. This could involve studying how individuals perceive themselves when viewing their mirrored virtual representation.