Optimisation of vocal music teaching strategy in colleges and universities based on neural network

This study optimises college vocal teaching by using a neural-network-driven system to address weak personalisation and delayed feedback.Tracking 50 students for 12 months (3,600 hours of audio and behavioural data), we built a model that analyses performance and generates individualised plans in real time.In a three-month intervention, students improved pitch accuracy by 18%, timbre richness by 21%, rhythm control by 24% and emotional expression by 27%.Long-term evaluation showed 35% higher sustained learning motivation than a control group, and overall satisfaction rose to 9.1/10 (+40%).Teachers also reported higher feedback effectiveness, rising from 65% to 88%.Results indicate that neural-network-based personalisation delivers measurable gains in technique, expressivity, and persistence while enhancing instructional efficiency.