Voice, vulnerability, and expressive growth: investigating AI anxiety and performance appraisal in voice arts education

As artificial intelligence (AI) voice technologies rapidly evolve, students training in vocal performance increasingly grapple with anxiety about their artistic identities and career trajectories. This study investigates how AI-related Anxiety (AIA) influences both self-perceived artistic expressiveness (SAE) and external evaluations of expressive performance (TAE). Anchored in the Vienna Integrated Model of Art Perception (VIMAP), and employing a revised version of the Expressive Power Scale, we conducted a mixed-methods study involving 414 vocal performance students and 414 independent raters across three Chinese universities. Quantitative analyses revealed that students with heightened AIA tended to undervalue their expressive capacities, even when third party ratings were favorable. Structural equation modeling indicated that AIA negatively predicted externally assessed expressiveness, and self-assessment had only a weak positive effect. In the qualitative findings, novel psychological themes were identified including emotional distancing and rediscovering one's creative agency in the face of AI. These results underline the significance of integrating both psychological skills and AI literacy into vocal performance training, while also ensuring students can retain their artistic identity in a potentially AI-driven context.