^^^ It's not that the fast-moving air conducts sound worse; there is no sonic boom. You could have an arbitrarily small gradient over a large space and this would be the case.
You could ask what the maximum gradient is before you get a sonic boom.... I would guess the characteristic length would depend on both the fluid properties of air like the density, *and* the size of the object creating the sonic boom. Not sure though... you could probably find the answer in a fluid dynamics book...