Automatic Computer Composition for Piano Music via Deep Learning and Blockchain Technology

Abstract

The aim of the current work is to examine how deep learning and blockchain technology may be used to create piano music automatically. First, blockchain technology's Ethernet Proof of Authority (POA) consensus method achieves distributed consensus across the network's nodes as a whole. This approach is well suited for the piano music alliance chain since it has an effective consensus efficiency and authentication mechanism. The automatic music generating model is built using these four neural networks after studying the properties of the recurrent neural network, Long and Short-Term Memory network, convolutional neural network (CNN), and multi-column CNN. The number of weighted parameter changes and learning increase together with the function's iterations, according to tests. As a result, this method can greatly improve the accuracy of the model for music creation. Additionally, the loss value of the loss function constantly falls as the number of iterations rises. Moreover, the methodology put out by other academics takes close to 2.3 seconds to process 1,000 pieces of data from piano scores, whereas the blockchain approach employed in this experiment takes only 1.25 seconds. Therefore, processing data from piano scores by computer using the blockchain concept is highly efficient. Hence, the current work holds significant implications for advancing the intelligence level within the realm of piano composition.