Research on the Style Transfer Fusion Model of Piano Improvisation Accompaniment Driven by Artificial Intelligence

This paper proposes a style transfer fusion model for piano improvisation accompaniment driven by artificial intelligence. The model combines CNN and RNN, extracts audio features in piano improvisation accompaniment through CNN, and uses RNN to process the time dependency in audio data, and finally realizes the fusion of style transfer and improvisation accompaniment generation. Different from the traditional piano improvisation accompaniment generation method, the model in this paper can adaptively learn the music features of different styles through deep learning technology to generate piano improvisation accompaniment that meets user needs. Experimental results show that the model performs well in multiple style transfer tasks, and can generate expressive improvisation accompaniment while ensuring style consistency, with an accuracy rate of 92%, which is 15% higher than the traditional style transfer method. The innovation of this model lies in the combination of style transfer and improvisation accompaniment generation, which improves the diversity and creative freedom of generated accompaniment. In addition, the model has strong real-time performance and can be applied to real-time music creation and performance.