With the continuous development of vocal music education, traditional teaching methods have gradually failed to meet the needs of personalized and efficient learning. This study proposes a learning path recommendation model for vocal skills based on knowledge graph and Dijkstra’s algorithm, which aims to provide a personalized learning path planning method for vocal education. First, the study constructed a knowledge graph of vocal music discipline containing 2245 knowledge points and 12 kinds of relationships, which covers the course content, knowledge points and their interrelationships. On this basis, Dijkstra’s algorithm was used to solve the shortest learning path from the mastered knowledge points to the target knowledge points. The experimental results show that the learning path recommended based on the model can effectively improve students’ learning efficiency. In the experimental group, the students’ vocal skill scores increased from 5.56 to 8.14, with an improvement of 46.4%. Compared with the control group, the score of the experimental group was significantly higher, and the personalized features of the path recommendation significantly improved the learners’ skill mastery. The study suggests that the learning path recommendation combining knowledge graph and Dijkstra’s algorithm can provide students with more accurate and efficient learning guidance.
