by james » Tue Jun 13, 2017 8:11 pm
The difference between voice/wind and guitar is very subtle.
For a signal that us truly monophonic (only one note) the pitch-detection algorithm is tuned slightly differently. Voice/wind or anything that can't produce two notes at the same time can often get better results in this mode.
If there is a change of more than one note (i.e., stringed instruments), the other modes take account of this and try to make a better decision about the note played even if there are other notes sounding at the same time.
The differences are mainly to do with notes being mis-detected, or being detected in the wrong octave. It really is fine-tuning of an algorithm that works almost perfectly most of the time in any case.