This blog will track the progress of Tim Kang and Michael Perry's CSE 190 project. Our project is on applying OCR to sheet music (project proposal here). Our goal is to eventually be able to read handwritten or badly deformed printed sheets, but we will start with well-formed printed sheets in order to get everything working. For this project, we are going to stick with sheet music written in normal western notation; none of that hokey stuff.
For the people with less experience, Music OCR consists of several steps:
-- preprocessing -> staff recognition/removal -> symbol segmentation/classification
after of which the sheet music can be reconstructed in a format such as MusicXML or MIDI and read by a scorewriter program such as the commercial Finale or the open-source MuseScore.
No comments:
Post a Comment