The algorithmic composition for music copyright protection under deep learning and blockchain

Abstract

To strengthen music copyright protection effectively, a new deep learning neural network music composition neural network (MCNN) is proposed. The probability distribution of LSTM generation is adjusted by constructing a reasonable reward function. Music theory rules are used to constrain the generated music style to realize the intelligent generation of specific music style. Then, the digital music copyright protection system based on blockchain is constructed from three perspectives of confirming right, using right, and protecting right. The validity of the model is further verified by relevant data. The results show that the composition algorithm based on deep learning can realize music creation, and the qualified rate reaches 95.11%. Compared with the composition algorithm in the latest study, the model achieves 62.4 percent satisfaction with subjective samples and a recognition rate of 75.6 percent for musical sentiment classification. It is proved that the music copyright protection model based on block chain can ensure that the copyright owners of works obtain corresponding economic benefits from various distribution channels, which is helpful to build a harmonious music market environment. In short, the innovation of this study is reflected in that it fills in the gap of detailed comparative study of the differences in the application of different models, realizes the framework of music copyright protection system, and provides convenient conditions for composers.