If the current transfer ratio is C, this means that when you pass a current I through the emitting side and apply whatever voltage is specified between the collector and emitter of the receiving side, you will get a collector current of C * I. Current transfer ratio is typically specified as a percentage. For the circuit I suggested, the CTR is not critical unless you are using low values of R. What values of R will you be using?

Using this calculator (http://sim.okawa-denshi.jp/en/CRtool.php), I played around with some target cut-off frequencies. With a cut-off frequency of exactly 30 Hz, the Bode diagram showed about 6 dB of attenuation at 30 Hz. At Fc=100 Hz, there was almost no attenuation at all shown at 30 Hz, and somewhere in the neighborhood of 30 dB of attenuation at 3 kHz (the PWM frequency you recommended). That seems like a pretty good outcome.

The potentiometer that I'm replacing is 22k, so each of the resistors will be 11k. The corresponding capacitor value to 11k, for Fc=100 Hz, is 1.4 uF. Using the closest standard capacitor value of 1.5 uF, we get Fc = 96 Hz. Or should I have been using 22k as the resistor value instead of 11k? The capacitor is only short-circuiting one of the resistors, so I think 11k is the correct value.

EDIT: The calculators on that site seem to disagree. If I use the calculator that takes a cutoff frequency as a parameter and work backwards, I get a cap value of 1.5 uF. But if I put 11 kOhm and 1.5 uF into the calculator that works forwards to derive a cutoff frequency, I get 9 Hz, not 99 Hz. I'm going back to the formula that was given earlier to calculate by hand.

EDIT #2: Yes. Hand calculations give 9.6 Hz. So the correct cap value is 0.15 uF, not 1.5 uF.