I was thinking it might be interesting to use an exciter as an informal test of a speaker cabinet. I would think a good cabinet would be about as ineffective a surface to put an exciter on as you could find. I'd imagine a thick sheet of lead would be pretty quiet. I was wondering how this might apply to a horn as well. I made a paper horn recently and put it in front of the tweeter on my bookshelf speaker. I heard very little change. The sound mostly goes right through the paper. Like a cabinet, a horn should block sound and be very difficult to structurally excite with sound, or so it would seem. I remember reading an article about the JBL 308P and the reviewer was worried about how thin the plastic was on the waveguide. So he added a bunch of epoxy to the back side to make it thicker and heavier. The result was... no audible change as far as he could tell. Years ago I added a bunch of rope caulk to the back of the thin plastic horns on my Kipsch RF-7s. I also heard no obvious change.
So if a thin plastic waveguide gives good results, why can't a thin plastic cabinet do the same? Maybe it's the lack of mass causing a lot of energy transfer directly in to the cabinet. If the woofer's frame was thick and heavy enough, and it had a big heavy mud magnet, and maybe if it used two opposed woofers back to back, the thin plastic wouldn't matter so much.
I made a wooden horn once that would vibrate in way that was very easy to feel when I bolted a lightweight 2" wide band driver to it. When I stuck a big heavy compression driver on it playing the same range of frequencies I could feel no vibrations, and could see a much less ragged measured response when smoothing was turned off. With 1/3 octave smoothing it looked about the same. I didn't have the ability to do time analysis on it back then, just frequency response. Also tried a piezo compression driver made by Panasonic, I think, that could reach down to 800 Hz. That thing sounded so shimmery and ethereal, reminding me of a fiery diamond with a spot light on it, if you could compare sight to sound. It had an intriguing response that was fairly smooth but with a number of very deep, narrow notches across the response, kind of like what you see when people do FIR room correction and just chop off all the narrow peaks in response, leaving only the narrow dips.