Jean-Pierre Briot, Gaëtan Hadjeres, François Pachet

This book is a survey and an analysis of different ways of using deeplearning (deep artificial neural networks) to generate musical content. Atfirst, we propose a methodology based on four dimensions for our analysis: -objective – What musical content is to be generated? (e.g., melody,accompaniment…); – representation – What are the information formats used forthe corpus and for the expected generated output? (e.g., MIDI, piano roll,text…); – architecture – What type of deep neural network is to be used?(e.g., recurrent network, autoencoder, generative adversarial networks…); -strategy – How to model and control the process of generation (e.g., directfeedforward, sampling, unit selection…). For each dimension, we conduct acomparative analysis of various models and techniques. For the strategydimension, we propose some tentative typology of possible approaches andmechanisms. This classification is bottom-up, based on the analysis of manyexisting deep-learning based systems for music generation, which are describedin this book. The last part of the book includes discussion and prospects.