Open hbitteur opened 7 years ago
Boundig boxes for numerator and denominator strings can be reconstructed by MuseScore. Its more expensive to do this for single glyphs.
Perhaps we can forget about single digit extraction, provided that we address the most common 2-digit numbers found in time signature (say at least 10, 12, 16)?
If we except CommonTime and CutTime, a time signature is composed of 2 numbers (numerator and denominator).
Should we consider a time signature (say 8/12, assuming this can exist) as one global time symbol (combo, such as "8/12") or as two time number symbols (numerator, denominator such as "8" and "12") or as two sequences of time digit symbols (such as {"8"} and {"1", "2"})?
We can handle a limited set of predefined combos. Some are already in OmrShape class, perhaps we could a few more,but we can't address all possible combos. We have to be able to recognize individual time digits. Even for a combo, could MuseScore generate number symbols (one for num, one for den)? Perhaps also single digit symbols? The more symbols, the better (even if they overlap).