Conditional Semantic Music Generation in a Context of VR Project "Graphs in Harmony Learning"

The article proposes a perspective on the use of generative artificial models in a context of the VR project "Graphs in harmony learning". The usage of LSTM, ConvLSTM and conditional GAN with convolutional 1D layers for semantic music generation is discussed. The efficiency of the novel data encoding scheme, along with the design patterns based on the system of graphs, are shown.