honest framing: YIN works best on monophonic content — a single voice or instrument. polyphonic content produces unreliable results; the tool reports the dominant pitch which may not always be the intended target. low-confidence regions are drawn as faint dots and excluded from the median / tuning detection. the source profile presets constrain the lag search to a sensible band (male voice 70-350 Hz, female 140-700, full voice 70-1000, instrument 60-2000) — pick the right one for your material.