This paper focuses on the field of higher music education, designing a digital classroom teaching system comprising three core modules: piano simulation, score editing, and performance demonstration. By incorporating MIDI audio technology, it details the system architecture and API functions, providing teachers and students with powerful tools for digital music composition and performance. Furthermore, it innovatively constructs a membership function model based on an A-V two-dimensional emotional coordinate system, enabling the quantitative recognition and intelligent classification of diverse musical emotions. To validate the effectiveness of the emotion retrieval model, this study constructed an experimental corpus comprising 17,834 songs and 102 emotion tags sourced from Apple Music. The results show that the membership function model proposed in this paper is significantly better than the mainstream algorithm in two key indicators, P@N and NDCG@N, with the highest P@N value reaching 0.892 in the Top40 results and NDCG@N 0.878 in the Top5 results, which proves its superiority in retrieval accuracy and sorting quality. Finally, an application survey conducted among 104 vocal music students at a university revealed that over 75% of students favored and endorsed the digital teaching model centered on this technology. Among the 8 professional instructors, 75% observed significant improvements in students’ creative abilities and learning motivation. This research not only achieved the digital transformation of traditional music teaching methods through its technical architecture but also demonstrated its technological advancement and pedagogical practicality through authentic experimental data and application surveys.
