Generating lyrics with variational autoencoder and multi-modal artist embeddings | Read Paper on Bytez