My name is Jeevan Thomas Koshy.I am a Second year Undergraduate in the Indian Institute of Technology,Madras.I have been practicing piano for 5 years now .I am also a part of a Church Choir for many years. I have experience working with computer vision. I have done a project which implements air writing with the help of a depth camera. I am extremely interested in working on something related to music. I want to create an OMR for Musescore in this summer as a part of GSOC 2016. I have installed Musescore and checked the existing OMR which was powered by Audiveris . I was suggesting on creating something similar for Musescore . I also have some other features in mind which can be used in OMR like separating a song into various parts .(so that the choir members can use that for learning).Please guide me on what to do.
First read https://musescore.org/en/developers-handbook/google-summer-code/steps-toward-successful-application
Then make sure you can compile MuseScore on your system, use MuseScore and take a look to our bug list https://musescore.org/en/project/issues/musescore to get familiar with the code base and concepts (measure, staff, voice, part etc...)
Currently, MuseScore uses audiveris running as a webservice to offer "Import PDF" feature. This is completely separated from MuseScore.
Audiveris represents years of work by his author Hervé Bitteur, I'm a bit doubtful anyone can implement something comparable over a summer :)
MuseScore itself has some code to open a PDF and try to guess how many staves they are per pages, how many measure per system, and what is the space between two staff lines in order to prepare a skeleton score and make it easier for people to transcribe. It's not a full OMR though, and it's not compiled by default.
Another "OMR" project could be the ability to recognize "in real time" handwritten music. This is not in the idea list, but it's maybe more aligned with your previous experience. They are many commercial implementation of this idea in mobile apps (staffpad, notateme, myscript, myMusicScore,...). I'm not aware of any open source implementation. Of course, this project is also a lot of work, especially if nothing exists already, and would probably be mainly exploratory... Still, it would be interesting to see how these concepts fit or don't fit with MuseScore current architecture.
The part separation idea is probably better focused. I'm not exactly sure what would be the input and output in your proposal. As you may know, MuseScore already support linked parts https://musescore.org/en/handbook/part-extraction-0 at the instrument level. From our discussion on IRC, I gather you are talking about voices https://musescore.org/en/handbook/voices-0 and this idea here https://musescore.org/en/developers-handbook/google-summer-code/ideas-2016#extract-part-from-voices
2016-03-11 8:01 GMT+04:00 Jeevan <[hidden email]>:
Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
Mscore-developer mailing list
|Free forum by Nabble||Edit this page|