Where the voice turns unpitched, or the detector loses it for a moment mid-note, they keep the voiced grains' alignment and return to the input's timing gradually (at once where it is quiet, or below ...
Some results have been hidden because they may be inaccessible to you
