Research on Music Creation System Based on Computer Artificial Intelligence Technology

According to the structure of sound type library, this paper gives the realization method of multi-characteristic music creation system. Firstly, the whole architecture of music creation is realized by using the modules of computer, memory card, display, computer and sound. An effective multidimensional audio timing modeling method is established by constructing a deep learning method based on long-term memory network based on recurrent neural network and combining with supervision mechanism. Through this model, the music style of the famous singer Justin is generated. The algorithm was compared to Justin's music library, which had a maximum similarity of 98.25%, with 84.15% of the works receiving expert reviews.