For DC, it makes no sense to speak about characteristic impedance; that's by definition the ratio of voltage and current in a transmission line of a sine of given frequency, and 0 is really a cornercase of "frequency".
For anything on a PCB, thinking about the characteristic impedance of frequencies in the audio range makes limited to no sense: for the characteristic impedance to be a useful measure at all, the described piece of transmission line needs to be a significant part of a wavelength long. And with wavelengths above 10 km, that's never the case. (It becomes very relevant for things like power distribution grids and old-school telegraph lines, hence the Telegrapher's equation.)
But sure, for an impossibly large PCB, you could calculate, simulate or measure the characteristic impedance. But it will certainly not be that high! Why would it? That would mean terrible things for the power transport.
Realistically, if you put in typical per-unit-length capacity and inductivity of, say, a 1 mm wide copper trace over 1.6 mm typical FR-4, I'd expect a relatively constant Z' of maybe 40 Ω for low frequencies (we ignore coupling into everything else in a large radius); but the losses of course would dominate the picture, because, again, characteristic impedance is only relevant for things that are wavelength-comparable or larger, and you really don't want the resistance of 10 km of 1 mm wide 35 µm thick copper.
