The DSP depends upon knowing how the speaker is going to behave, shaping the output of the amp to take that predicted behavior and make the actual response flat. The problem is that much above 100 Hz, and there becomes too many variables, happening too fast, that it's impossible to flatten the response of the speaker with the same level of accuracy which is possible with a much slower moving target.
Impossible for how many more years? How fast does the processing power of DSPs increase, and how much processing power is needed to take care of the audible range? Those would be interesting figures to know. And as far as room correction software goes, you'd think that SOMEONE out there offering multi-thousand dollar equipment would be able to include enough processing power. I keep dreaming, anyway.
