ORPHEUS
Optical Recognition Parser of Harmonization with Edition Utilities for Scores
The project aims to semi-automate the digitization of printed musical scores, reducing manual effort and enhancing digital music-notation workflows. Success is assessed by the system’s ability to convert notation into an editable digital format, enabling users to refine results within a single environment, and by its reliable operation under typical conditions. Additionally, the recognition model is tested on a representative dataset, with a goal of achieving at least 60% symbol accuracy.
Transcribing printed sheet music into digital form remains largely manual and fragmented. Musicians and educators often depend on slow, error-prone workflows that involve switching between multiple tools for scanning, recognising, correcting, and exporting notation. Poor scans, annotations, and print artefacts further hinder recognition, making accurate transcription challenging without extensive manual effort. Existing Optical Music Recognition (OMR) solutions tend to be unreliable with real-world materials, inconsistent across notation styles, or confined within costly, closed ecosystems. Consequently, users encounter unnecessary effort, decreased productivity, and limited access to modern digital-notation workflows. This project aims to solve these problems by providing a more accessible, consistent, and integrated method for digitising and enhancing musical scores.
- Desktop application for viewing, editing, and exporting digital music scores
- OMR component for processing score images and generating structured notation
- Mobile app for capturing or selecting score images and sending them to the system
- Toolkit for training, testing, and validating the recognition model
Musicians, educators, students, and cultural institutions benefit from faster, more accessible score digitisation and smoother workflows for editing, teaching, and modernising musical materials.
