In contemporary society, stress has become a significant concern, further exacerbated by the COVID-19 pandemic, which has heightened awareness of mental health issues. Smartphones now play a crucial role in mental health management, with music-based stress-relief applications gaining popularity for their effectiveness in stress reduction. However, most of these applications employ a unidirectional playback model, lacking interactive features that cater to diverse user preferences, lifestyles, and emotional needs. This study utilizes task analysis to examine the influence of user involvement levels on interactions with music-based stress-relief applications. Three existing applications were analyzed, with 32 college and graduate students participating in the experiment. Participants were classified into high-involvement (n = 15) and low-involvement (n = 17) groups based on the Personal Involvement Inventory (PII). Objective performance metrics were obtained through task completion times, while post-experiment assessments included the System Usability Scale (SUS), NASA-TLX workload evaluation, and the Questionnaire for User Interaction Satisfaction (QUIS), supplemented by semi-structured interviews. Findings revealed significant performance differences across applications, particularly in custom playlist creation. High-involvement users preferred minimalist designs for quick access, whereas low-involvement users favored feature-rich applications. Based on these insights, three design recommendations are proposed: (1) personalized settings tailored to user involvement levels, (2) enhanced usability through clear guidance, and (3) a balanced approach between feature integration and streamlined navigation to optimize user experience.