AI Voice Features and Emotional Self-Dominance: Shaping Student Engagement in Instructional Video Learning

Guided by the Valence-Arousal-Dominance (VAD) framework, this study examined how AI-generated voice features in instructional videos influence secondary school students' engagement intentions, mediated by emotional self-dominance (a learner's perceived control over their emotional states). A total of 340 Chinese Grade 8–9 students engaged in mathematical problem-solving tasks while viewing videos that systematically manipulated three voice features: gender (male/female), pitch (low/medium/high) and prosodic prominence (present/absent of emphasising key information). Structural equation modelling revealed significant direct effects of high pitch and prominence on positive engagement intentions toward video-based instruction. Furthermore, self-dominance significantly mediated the relationship between gendered voices and learners' engagement intention levels, with female voices indirectly enhancing student engagement. These findings enhance our understanding of how voice features shape learners' emotional self-dominance and engagement intentions, providing practical guidance for designing instructional multimedia resources in AI-powered educational systems.