My Open MusicXML Library

Dear Music Notation Community Group:
       Hello! I've been working on my open MusicXML library, and I'd like to introduce it to you.
       There are two parts in the library. The first part is parallel to IMSLP. Each PDF from IMSLP is converted to MusicXML by two Music OCR software.. The accuracy for PhotoScore is 83.80%, and for SharpEye it's 89.25% (V. Padilla, et.al., 2015). Most of those errors are in rhythm. Although the accuracy of raw outputs seems to be poor, there are various of ways to improve accuracy, such as voting from different files (V. Padilla, et.al., 2015), using multiple recognizers (I. Knopke and D. Byrd, 2007) and musicology analysis in rhythm and chord (many people have worked on it). I've tried some of them, but I believe it's the users' rights to choose which and how. There're sufficient raw files on my database, so users can use any one of those fancy methods. And when new methods are developed, users can adapt new ways because they have raw materials. Or, they can manually pick out best ones and use them without modification.
       The second part of the library is converted from classical midis which are inputted by human. Strictly speaking, they are not real MusicXML, but they're more accurate. This part is for people doing small-scale musicology research.
       I have converted the complete works by Sergei Prokofiev and Alexander Scriabin. Most works don't have an existing analytical format. In the future, I'm planning to set up a community which both contributes new files and substitute old ones with better ones.
       The site is www.mscorelib.com  . I'd appreciate it if you can add it to the MusicXML site list (http://www.musicxml.com/music-in-musicxml/).
       Thank you very much!
                                                                                                                                                                                                                                              Best regards,
                                                                                                                                                                                                                                              Fang Shichao

Received on Friday, 15 September 2017 11:41:22 UTC