The current tsunami of deep learning has already conquered new areas, such as the generation of creative content (images, music, text). The motivation is in using the capacity of modern deep learning architectures and associated training and generation techniques to automatically learn styles from arbitrary corpora and then to generate samples from the estimated distribution, with some degree of control over the generation. In this article, we analyze the use of autoencoder architectures and how their ability for compressing information turns out to be an interesting source for generation of music. Autoencoders are good at representation learning, that is at extracting a compressed and abstract representation (a set of latent variables) common to the set of training examples. By choosing various instances of this abstract representation (i.e., by sampling the latent variables), we may efficiently generate various instances within the style which has been learnt. Furthermore, we may use various approaches for controlling the generation, such as interpolation, attribute vector arithmetics, recursion and objective optimization, as will be illustrated by various examples. Before concluding the article, we will discuss some limitations of autoencoders, introduce the concept of variational autoencoders and briefly compare their respective merits and limitations for generating music.