Yeah sweeps are no good. Though Praxis has this thing called a 'chirp' which sounds very much like a sweep, but supposedly there's some real fancy math going on with it (and I assume something in the signal that don't recognize as being in there along with the sweep) but it gives better results than I've gotten in the past with standard pink noise or MLS techniques. Anyway....
The standard way these days for measuring frequency response is by using a maximum length sequence technique. The fancy computer program that uses it (Laud, MLSSA, Praxis, LEAP, etc) gets you an impulse response, and you visually inspect the impulse response for any ringing that may occur as a result of room boundaries (using the speed of sound, the distance of the boundary to the floor and the mic, and some math you get the expected arrival time of the first reflection). Anyway, you look for that first reflection glitch in your impulse response, window the time you are going to FFT to just before the start of the impulse to just before the start of the first reflection, and FFT that. Only problem is that this limits you to a few hundred Hz on the low end. Good thing is most speakers are really really linear <300Hz or so so you can just close-mic the speaker to measure frequency response and let the direct sound from the cone swamp out the room reflections (you want the mic 1/2" from the cone in such a test).
Anyway, that's the way I do things when designing a speaker. I haven't had much time for the hobby since I graduated and started working a year ago, but I'll dig up an old design or two to show off.