This repository contains the instructions to downloading the dataset used in Optical Music Recognition with Convolutional Sequence-to-Sequence Models. The dataset contains approximately 17.000 monophonic scores, divided over a train, validation, and evaluation set.
Only the public keys to scores on MuseScore are included in this repository. To obtain the data, you need a MuseScore API key. More information about obtaining a key and the API can be found here. An example download script for python 3 is included in the scripts folder.