The plates with the holes act as magnetic field guides (a "pole piece" in a loudspeaker driver) to concentrate the field where the diaphragm coil sits (and increase the sensitivity). The field lines attracted by the guides follow the material around the holes within the guide material to exit the guide sides and get back to the magnet. If you want to start carving at the holes I'd suggest maintaining the wall thickness between the holes - I imagine excessive field concentration (i.e. closer to saturation) makes the guides less effective. I don't imagine an acoustic benefit to be had by modifying this aspect of the driver, although deburring or rounding the edges of the (probably) stamped holes might make me feel better about what little flow is there. Make sure you scrape all the metal bits stuck off the magnets afterward. Below the first natural frequency (everywhere <1kHz in my approach) the operating shape is basically the same as the first mode shape (see above). If you wanted to damp 100ish Hz by damping the center of the center magnet you'd be damping potentially everything up to around 2kHz (and other modes - 3.2kHz included - that show antinodes in the center). That sounds like an interesting experiment. Today I still maintain my expectation that the 100Hz rise is a mode arising from the interaction of the diaphragm "Vas", and the mass, stiffness, and terminal impedance on both the cup and pad sides of the diaphragm. The diaphragm itself appears to resonate beginning at 1kHz, and the extra mass and lower compliance of the "stuff" around it would generate the lower frequency responses. My model doesn't incorporate that at the moment, but if someone wants to suggest an equivalent mechanical system of springs and masses I could couple that with the diaphragm and see what happens.