Just to confirm it's a 2-way with a little "super" tweeter stuck on top. There is no low pass filter on the main tweeter--it just rolls off on its own. I'm still not exactly sure what you're looking for--it's obvious the response deviations in the 800 - 1500 Hz area are embodied in the woofer measurement, but here's a plot that shows the overall frequency response and the filtered response of each driver. The cabinet resonance may be contributing a little to the bump, along with inadequate baffle step compensation and probably also a resonance from the junction of the woofer cone and surround.
View attachment 172945
First of all thank you. I think I understood the topology.
As far as I can see there two issues. One is the peaking of the low frequency driver at around 1200Hz. I do not think this is caused by the 2Pi to 4Pi conversion (the baffle step). Correcting it should be possible by integrating low Q filter to the LP filter of the crossover.
The second issue is not solvable. I think the tweeter is about 2dB too efficient. I think the HP filter is not optimised either, causing the dip at around 3kHz.
Extra complexity to the crossover will add cost but reducing complexity, hence saving may well be possible by running a simulation. Filter simulation is pretty cheap these days.
Finally, has anyone tried to audition the speaker without the super-tweeter active? As far as I can see the red curve is pretty OK for a cheap speaker. It is flat to 15kHz! Swapping the cost of that driver with more complex crossover may even save the manufacturer money while offering a better speaker.