Imho, it would be easier to return the pre-delay which has been added to the driving function by a particular processing step, e.g. fractional delay, wfs prefilter, focused source, plane wave, etc..
Furthermore, I would prefer that t=0 corresponds to the t=0 of the virtual source, i.e. where the sound "starts" to radiate from a point/focused source or when the plane wave passed its point of zero phase.
This is a follow up on #59.
The main goal was fomulated by @fietew as: