[Back]

J. Calvo-Zaragoza, J. Hajič, A. Pacha, I. Fujinaga:
"Optical Music Recognition for Dummies";
Keynote Lecture: 19th International Society for Music Information Retrieval Conference, Paris, France; 2018-09-23 - 2018-09-27.

Optical Music Recognition (OMR) is a field of research that investigates how to computationally decode music notation in documents. As most musical compositions in the Western tradition have been written rather than recorded, bringing this music into the digital domain can significantly diversify the sources for MIR, digital musicology, and more broadly lower the costs of introducing previously unheard works to audiences worldwide. While OMR has been regarded as a largely unsolved problem, this situation has recently shifted: new large-scale datasets and tools have been released, methods based on deep learning are successfully dealing with musical symbol detection and partial end-to-end recognition, and applications of OMR such as retrieval have started migrating from article introductions to the Results sections.

Our tutorial will present this new and rather exciting state of the art in OMR. We will demonstrate recent methods and results, introduce the audience to the tools and datasets used to achieve them, and showcase the opportunities for using OMR. Finally, we will introduce the current challenges in OMR.

After the tutorial, the participants should be familiar with state-of-the-art OMR research, and should be able to start using existing tools to integrate OMR into their own work, whether in MIR or (digital) musicology. For those interested in working on OMR themselves, the tutorial should provide a head start. The tutorial will be hands-on: if you wish to get the most out of it, be ready to follow jupyter notebooks.

Optical Music Recognition, Tutorial

https://publik.tuwien.ac.at/files/publik_282963.pdf