With the intensification of social competition and the diversification of campus life, college students are confronted with increasingly prominent psychological challenges such as academic pressure and interpersonal conflicts. The accumulation of negative emotions and anxiety issues has become a key factor affecting their physical and mental health. Against this backdrop, developing low-cost and easily scalable emotional intervention methods holds significant practical value. As an intervention form integrating emotional transmission and cognitive regulation functions, the actual healing effects of different audio reading modes have not been systematically verified. This study recruited 240 college students as participants and adopted a 3 (empathic human voice reading/AI-synthesized voice reading/self-reading) × 2 (universal emotional content/targeted content) mixed experimental design. The State–Trait Anxiety Inventory (STAI) was used to assess changes in anxiety levels before and after the intervention, and data were analyzed using multiple statistical methods. The results showed that there were no significant differences in state anxiety, trait anxiety levels, or demographic characteristics among the groups before the intervention, indicating balanced baseline conditions. After the intervention, the empathic human voice reading group exhibited significantly better anxiety reduction effects than the AI-synthesized voice reading group, while the self-reading group showed intermediate effects with no significant differences from the other two groups. Targeted content focusing on core concerns of college students (such as Introductionacademic pressure and interpersonal growth) generally demonstrated superior healing effects compared to universal emotional content, and this advantage was not affected by the reading mode. No significant interaction effect was found between the two variables. This study reveals that the empathic expression of professional human voices and content design tailored to college students’ actual needs are key factors in enhancing the healing efficacy of audio reading. Self-reading exerts an auxiliary healing effect through active cognitive engagement, while AI-synthesized voice can serve as a low-cost supplementary solution. The findings provide new insights for universities to construct a diversified mental health intervention system and offer empirical references for the emotional optimization of AI mental health products.
Yang et al. (Mon,) studied this question.