Modeling Method of Intelligent Scoring for Solfeggio Training Based on Deep Learning

Abstract

In the solfeggio training, melody model singing is an important part of auditory training. Reasonable and effective model singing training is of great significance to train the singer’s hearing. So, the level of model singing also reflects the level of the singer’s solfeggio to a certain extent. The correct model singing score will help students to master their own real melody model singing level and help students to better improve their solfeggio skills. However, the melody model singing training scoring system that has been developed has low accuracy and needs to be improved. In order to improve the accuracy of the intelligent scoring results of melody model singing training, this paper used deep learning technology to study the intelligent scoring method of melody model singing training. This paper first extracted the features of songs, then collected a large number of song features, trained the existing feature data, and established an intelligent scoring model for melody model singing training. Then, the students’ model songs and standard songs were used as two inputs, and the similarity of the characteristics of the two songs was analyzed; then, the score results of the model songs were obtained. The validity of this research should be verified by a comparative experiment. The research results showed that the melody model singing training intelligent scoring model, established by deep learning technology, has a higher accuracy rate, and the accuracy rate was increased by 6.82% compared with the original scoring model, indicating the research has practical significance.