Yup. Noctua, for example, spends a whole lot of money working to reduce the perceived noise in their sound profile. While Noctua probably spends the most, you will also note toothlike trailing edges in Nidec Gentle Typhoons. Those are designed to shepherd their SPL into more pleasant tones.
This is all you can do. You can't fight physics.
Measuring the SPL of fans on rads, behind filters and open but on a box runs into the problem of mixing box resonance. When I measure the RPM and SPL of a fan in open air, I do it using one of three foam-lined stands that do not contribute any noise from vibration. I am not capable of precision woodwork, so all of my mixing boxes are cardboard, and not suitable for SPL measurement. Also, my methodology has changed as I have come to understand better the workings of the anemometer. Cardboard allowed me to make changes quickly.
The RPM's of the fans differ, depending on what they're mounted on. I have been recording this but not posting the data because there wasn't room on the charts. But with my editors telling me that the main chart is too dense, I'm figuring that I can now make charts which show the RPM of each state -- free air (master chart), pushing on a rad, pulling on a filter, and on the box, not formally obstructed (I'm not convinced a mixing box is not obstructing in some way). I'll do that next time. Some fans show increased RPM with resistance, some fans lower RPM. Fascinating for a fan geek.
Perhaps now I have arrived at a sort of climax design, so that making a sound box out of wood instead of cardboard would make sense. A fan would still resonate, contributing to the SPL unless there is a vibration-attenuating way to mount the fans. I'm open to suggestions on that -- because fans indeed sound different when they are behind a filter or mounted on a rad. Your thoughts will not go unnoticed.