{"references": ["Curtis Hawthorne, Andriy Stasyuk, Adam Roberts, Ian Simon, Cheng-Zhi Anna Huang, Sander Dieleman, Erich Elsen, Jesse Engel, and Douglas Eck. \"Enabling Factorized Piano Music Modeling and Generation with the MAESTRO Dataset.\" In International Conference on Learning Representations, 2019."]}

269.719 Images of size 128x128px from the MAESTRO V2.0.0 dataset available at the Magenta team webpage. The MIDI performances were binarized, and splitted into sections of 128 timesteps. Each section was transformed to a grayscale image.