Open rebmizrahi opened 11 months ago
To clarify: even when original MIDI and Music XML files have the correct amount of voices (e.g., 4), the secondary MIDI file resulting from automatic Music XML -> MIDI conversion has the voices collapsed into one. So, the MIDI file resulting from conversion is misencoded (the problem is not in the jSymbolic feature extraction, it is in the conversion). This will therefore affect all features that consider distinct voices.
jSymbolic recognizes n distinct voices in an XML file, but after converting it to MIDI it only recognizes 1. Cory noticed this specific bug but there may be other features affected.