Right now frames must be equi-distant and dictated by frame size for re-synthesis. We don't actually need this if the synthesis can re-interpolate, and would allow for less data when recombining files with different frame-sizes or when the combination is not "on the frame grid"