What would it be with these parameters:
Configuration
Driver type(s)
Frequency response and timbral characteristics
Sensitivity
SAF - a BIG one
And anything else you want to throw into the mix.
Just thinkin' here.
Depends on the budget and space.
Either way I like to make speakers respond as much like a pure voltage devices as possible, but not in the most traditional of ways. Generally I prefer anything but sealed to compensate for natural impedance without a crossover. And then the crossover takes some interesting modifications not typically seen.
When choosing drivers I have no reservation for any kind of mantra. If the spec's fit I am interested. Nice waterfall's are good to see. In general I like cone and dome. I have really enjoyed different planar types at different times, but feel like they don't have the full potential cone and dome done really well can have. I have yet to thoroughly jump into horns, but can say my interest in them would be linearity, not CD.
The QES/QMS/QTS does interesting things. Too much QES (lower number) and the speaker will be restricted a lot, especially for bass. With QMS it is funny because it's a mixed bag. While less of it makes a more voltage driven unit, you can also suffer from QES dominance it seems, with too little (higher number). QTS is the measure of the two, where QES is highly dominate. I haven't decided whether I prefer bass from a mid QTS woofer, or prefer to start with a lower QTS woofer that responds well to porting, then tune it up a little by lower QES (increasing the number).
I think higher BL is good.
Higher sensitivity is appreciated as I think low efficiency speakers are a pain to power, it's costly. I am not saying I would never have one, and the amps to back it up, it's just not my first approach.
Because of how I like to use drivers in the crossover I basically prefer higher inductance since there are attributes I want that come with it. I am able to negate the negative effects of it well enough.
Configuration... depends what you're building. I think 3 ways are pretty optimal. It is probably possible to do a good 2 way but I haven't been convinced just yet. While I am all about not going sealed, the frequency range of the midrange will be some of the reasoning on whether I want it sealed or not. If it's a long way from it's impedance spike/s then it's ok. And I actually am fine with using bipolar configurations as well. They are tricky and need tuning but I like their benefits for design of reducing the need for too many drivers and possible comb filtering.
I highly recommend fighting diffraction, it causes high frequency fatigue.
Box material is important. I prefer rigid construction. Aluminum is awesome, and a variety of other things beyond the average. Simple MDF works but isn't remotely close enough to deadened enough even with no-rez applied and braced; it can sound good but is limited ultimately. That is another reason bipolar can be useful, to circumvent needing more costly, heavy, box construction.
When it comes to treating the inside of the box I hate stuffing. IMO everyone is wrong to use more than the slightly puff you can see through. It kills sound. But almost ironically lining the walls seems to improve sound. The goal is to reduce return waves in the box, and reduce vibration. What you line the walls with does add a touch of dampening but doesn't seem to have as negative of consequences as stuffing does that over dampens the speaker. I think nearly all speakers sound very over dampened which greatly hinders believability as there is no believable reverberation in the instruments you hear.
On and on and on... this is why I haven't made a speaker project design just yet. Although I certainly am working on getting there.