Research

This work is a field study into user acceptability and perceived value of AI-based virtual singers in music production. A seven-part quantitative survey (N=378 participants representing the global music community who are practitioners/teachers) followed by a qualitative sentiment analysis that identified technical familiarity, simulation fidelity, cost efficiency, and functional/cultural utility to be key factors influencing adoption. Quantitative analysis validated four hypotheses: (1) Technical familiarity significantly correlates with acceptance (χ²=25.380, p<.001); (2) Experienced producers prioritize fault tolerance and economic cost (χ²=17.853, p<.001); (3) Simulation fidelity predicts expectations of future impact (r=.567 with “surprise”, r=.450 with “revolution potential”, p<.001); (4) Functional utility is valued over cultural applications (t=19.836, p<.001). The sentiment analysis of open-ended responses indicated that overall, respondents had mostly positive/neutral attitudes (52 percent positive and 29 percent negative, respectively), but criticism focused on continuing drawbacks in the expressiveness of emotions. The reliability studies showed strong scales in performance appraisal (Cronbach=.814) and lacked usability design options in editor features (Cronbach=.424). The findings reveal that, although AI virtual singers present transformative cost effectiveness (78 percent agreement) and workflow benefits, the use of AI virtual singers is limited by technical accessibility and emotional authenticity roadblocks. As we suggest, the optimization of technologies and, as a priority, the development of affective synthesis models should be done in an occupational-group-specific way. The study establishes empirical grounds for how to improve the AI voice tools and integrate them into creative processes.

Keywords: Artificial Intelligence; Singing Voice Synthesis; Virtual Singer; Technology Acceptance Model; User Experience; Music Technology; Affective Computing; Psychology of Music