Run it through the program and OCR it. Describe the process for the case studie. How to ensure quality (we need to do some type of testing)? What doesn't work?
Description
Status | Subtype | Assigned | Task | ||
---|---|---|---|---|---|
Resolved | Jopparn | T215269 FindingGLAMs media uploads | |||
Resolved | Alicia_Fagerving_WMSE | T217635 Upload audio files from Musikverket | |||
Resolved | Alicia_Fagerving_WMSE | T220071 Upload images from Musikverket | |||
Resolved | Alicia_Fagerving_WMSE | T221960 Upload sheet music from Musikverket | |||
Open | None | T223083 Add the sheet music in structured format |
Event Timeline
https://github.com/Audiveris/audiveris seems to be a useful tool. It opens the scanned sheet music and does something to it. According to the manual, it can export to MusicXML. Then there are tools to convert that to LilyPond format.
There's a long discussion about written music formats on Commons.
More reading material:
ON WIKIDATA:
- Property LilyPond notation.
- Property musical quotation or excerpt. It also uses LilyPond format, so what's the difference? Used e.g. here.
I think it's great if someone with knowledge on the subject matter can take care of this, yes.
The first thing to do is browse through the documentation – the links I posted, as well as anything else that looks relevant. What we learn from those will inform our decisions.
The main issues, as I see it now:
- The technical part of turning a scanned image / PDF to a structured format compatible with Wikimedia projects. How hard is it – can you achieve good enough results with one press of a button, or does it include a lot of manual work (by a musically-trained person). I assume the musical OCR will generate errors, as is expected with any OCR :) – but is it more "this doesn't make any sense at all, doing it by hand is actually faster" or "let's ask the community to help proofread those and catch errors"?
- Musical notation on Commons. I'm not even sure *what* our goal is here, so basically – what *can* we do? Examples of others having done it?
- Ditto about Wikidata.
We didn't upload anything, but the findings we did when researching this were included in the FindingGLAms white paper.
Removing task assignee due to inactivity, as this open task has been assigned for more than two years (see emails sent to assignee on May26 and Jun17, and T270544). Please assign this task to yourself again if you still realistically [plan to] work on this task - it would be very welcome!
(See https://www.mediawiki.org/wiki/Bug_management/Assignee_cleanup for tips how to best manage your individual work in Phabricator.)