Application of Artificial Intelligence and Speech Data System based on Music Internet Course Learning System

Abstract

Abstract Music education, as an important part of art education, can also take advantage of artificial intelligence technology to achieve more efficient personalized teaching. The direction of the application of artificial intelligence technology in music course learning is explored in this paper. Therefore, a music data technology interactive learning system is proposed. The music recognition module’s design utilizes a deep neural network model to model the complex problem of speech signal recognition. The encoder formula is obtained by representing the implicit layer feature vectors in the sample through a mathematical model. After encoding and decoding, as well as designing the activation function, the HMM algorithm is introduced to realize the application of DNN-HMM in acoustic modeling. Using a digital filter, the spectrum of the speech signal is smoothed, and the spectrogram is obtained by Fourier variation to visualize the representation of the speech frequency domain. The design of a music Internet teaching course is based on the method proposed in this paper. The melody recognition accuracy of the system is tested through simulation experiments, in which the distribution of auditory feature points of the piano ranges from 0.66 to 0.69. The distribution of rock music is above 0.7, and there is no overlap between the two audio datasets, which indicates that the system proposed in this paper has good recognition accuracy of audio features. Using the speech analysis module, the students’ music learning performance is analyzed. After the model designed in this paper to assist music learning, students’ music performance mean value is 4.397, and the control group’s performance is 3.565. The difference is 0.832. The system designed in this paper is more effective for music learning.