Abstract The arrival of the intelligent era makes intelligent music education a reality, which is a product of the combination of intelligent technology and music education. The article starts with the intelligent transformation of music education, analyzes the talent cultivation mode of music education, and establishes the intelligent classroom teaching mode of music education through the use of intelligent technology. Combined with the data of students’ online music learning behavior, a personalized music learning path model is established, and a nonlinear factor is introduced to optimize the SBPSO intelligent algorithm in order to realize the accurate recommendation of a personalized learning path. A DFN model for the automatic generation of music accompaniment was established based on the Flow model, and simulation analysis and teaching experiments were carried out to analyze the effectiveness of the above methods in music-intelligent classroom teaching. The convergence value of the NFSBPSO algorithm was in the range of 0.026~0.043, and the discrepancy between the mean value of the note length of the accompaniment generated by the DFN model and that of real data was only 0.35% and the total average score of the generated accompaniment was 7.0%. Accompaniment had an overall mean score of 7.23. The observation group scored 83.64 on their final music course grade, which was 11.46 points more than the control group’s average grade. Combining music education with intelligent algorithms can achieve intelligent transformations in music education and improve the quality of talent cultivation.