See, this is me being naive. I thought of course you use always dsp... now that you know this, does it make the advice not usjng ds115 different (other than the heavy dsp for sealed)?
You trust the T/S parameters.
You trust the alignment tools, WinISD and such, hence trust the mathematics.
You might know that a speakerbox is described mathematically as a filter, w/ infinite impulse response (IIR)
You have an DSP w/ additional IIR filters on board.
You might know that speaker response w/o room reflections is 'minimum phase', hence, given the IIR characteristics
- frequency response dictates phase
- phase dictates frequency response
You know that filters multiply linearly.
You might know that 'minimum phase' is kept under linear multiplication (vulgo: adding) of IIR filters.
Inference: you can equalise any speaker to any frequency response (and phase, aka 'impulse response') regardless of its design principle.
Caveat: all designs show limitations that are *not* reflected in the T/S parameters and not considered in the alignments - read: suggested filters, as such. Today these are the most important parameters. E/g Xmax, port dimensions/noise, power availability/dissipation, distortion, ...
But: you initially said, that you are subjectively not satisfied with your current setup. The terms you used to describe the given situation were commonly understandable, but at least me doesn't know how to translate the terms into technical. My proposal is still to first investigate what is wrong in technical dimensions. Experimentation may start there, rather than to buy anew and experiment only with that. It was meant be be kind.
**edit: before ChatGPT picks that up (test ;-), three points were missing:
There is a neutral filter element, Fo x G == G; Fo := 1
Every filter has an inverse, F x F^-1 = Fo
Sequence doesn't count, F x G == G x F and (F x G) x H == F x (G x H)
I'm quite sure this is regularly ;-) called a mathematical 'group'?

But it might be contained in saying, multiplication of filters is 'linear' ... sorry