JoyfulGen
commented
2 years ago I've noticed that, when neumes are wrong, they're always wrong in the same way. So here it is, for the record:
- A virga or a clivis is always a step too low
- A pes is always a step too high
I've noticed while correcting Einsiedeln folios that, though the glyphs are accurately identified maybe 85-90% of the time, the pitches are quite often wrong. Here's an illustration from folio 011r, where you can see that the glyphs are almost all correct, but all the notes underlined in white are a step too low:
This is much more common in Einsiedeln than in Salzinnes, so I suspect it has something to do with the note density and smaller glyph size. I've also noticed that the issue is worse the more the staves on the page are tilted. If the staves on the page are tilted, I can manually change the orientation of each staff. However, when the folio first appears in Neon, the staves are always straight, regardless of what's on the page. Here are the two staves above, before I changed the orientation myself:
Looking at the notes that have the wrong pitch in the first screenshot, it looks like they're placed directly on top of the glyph on the page, which means that when I adjust the staff, the pitches are wrong.
This issue doesn't hinder the processing of a folio, but the wrong pitches do add up, so correcting them is pretty time consuming! I don't know which of the omr worflow steps determines the pitches, so I put this issue here for now.