Optical Music Recognition from Multiple Sources

Pre- and post-processing software plus control scripts to improve output from optical-music-recognition (OMR) software by combining information from multiple sources. The system requires individual OMR recognizers to read score image files in TIFF format and to produce output files in MusicXML format. It has been tested with Capella-Scan, SharpEye, SmartScore and PhotoScore, but in principle could work with any OMR software which meets these basic requirements. Besides working with the output of multiple recognizers, the system is also able to combine information from multiple images of the same piece of music (including different editions) and to use parts and scores.
This project was funded by an AHRC research grant under the Big Data scheme as part of the Digital Transformations programme. The software uses the music21 and Gamera Python libraries and Sikuli script.