If you already know the pulse quotient, then you can just settle for measuring the peak voltage. The error will mainly depend on how fast the circuit switches, i.e. how well the curve follows the square wave you are switching with.

Trying to sample both the high and low values is a dead end, since you can't get your ADC to measure fast enough to catch the flanks, so you don't only get errors from rise and fall times but from not getting samples directly before and directly after the flanks.

The other alternative is of course to have a long enough filter constant that you measure the average voltage.