The purpose of this study is to compare academic results and psychological factors of influence in the context of the use of deep learning technologies. The experiment involved 238 respondents who were divided into two groups -- control and experimental. Students were tested for academic self-efficacy and well-being after taking the exam composition. Most of the students in the control group (n = 61) received satisfactory grades, and the index of academic results improved significantly. It was found that academic self-efficacy correlated much better with the results of the experimental group than with those of the control group. At the same time, the correlation between students' well-being is somewhat lower than the previous indicator but remains high. This study has practical value and is important for further research because it demonstrates the impact of the two factors studied on academic achievement in music in the context of using deep learning technology versus traditional education methods. The study results can be used by universities, music schools, and other educational institutions that are interested in developing the musical skills of their students as part of the global digitalization of education.