This is the data package that contains training data, training scripts and trained models presented in the paper "An OCR system for the Unified Northern Alphabet", by Niko Partanen and Michael Rießler, to be published in the proceedings of IWCLUL 2019. All scripts were harmonized to work in other environments.