100uF low ESR cap, 10R 0.5W resistor and another 100uF low ESR cap after it, both with a 100n chip ceranic across them. negative lead to go to the capacitor pad and then to the first capacitor pad, then connect to the ground plane.
I've discovered that different model of fans, even from the same manufacturer, have different noise (more than 20 dB difference between two 24 V, 80 x 80 x 25 mm fans from a top-rated Company (SunOn)).
SeanB' proposal works well, but I prefer to power the fan from unregulated DC, before voltage regulation, with an appropriate series resistor (to get the correct voltage across the fan) and a BIG electrolytic capacitor in parallel with the fan. The negative wire must be connected to ground at a suitable point, to be found by experiment, usually near the negative pin of the main supply's filter capacitor.
Noise seem to increase with current, so using an higher voltage fan sometimes helps.
My preferred solution (with dual rail supplies) is to power the fan from positive unregulated DC to negative unregulated DC, with two suitable series resistors, a big electrolytic in parallel with the fan. No connection to circuit ground.
Typical values (from +/- 24 V dc unregulated) are two 330 ohms / 2W resistor and a 470 uF capacitor, resulting in about 15 V driving a 24 V fan at a lower than nominal speed.
Best regards.