I feel like there is two points that people often miss with EQ:
1) only when the system behaves like a minimum phase system (smooth Excess group delay) that EQ can meaningfully correct a response, otherwise the phase and group delay will not be corrected and the ringing will continue to be audible.
2) steady state (regular in-room) measurements only represents what we hear below 100Hz.
Here are some examples of my room, we can see that EQ pretty much corrected a lot of the ringing below 100Hz.
View attachment 229509
And it's even visible in the Group Delay
View attachment 229510
This is because my Excess group delay looks something like this.
View attachment 229499
While i could correct for the dips in 30 to 40Hz they're very seat dependent so i left them be.
This is what the overall response looks like (1/12 oct smoothing) with EQ only below 100Hz, speakers are flat on-axis with smooth directivity. It sounded still muddy and colored, but with good sounding sub-bass.
View attachment 229500
We can see that there are still messiness below 1KHz, much of which is not very different from seat to seat. only below 300Hz where things really change dramatically between seats. Maybe we can use EQ to correct it? Lets have a look at Excess Group Delay.
View attachment 229503
Yeah looks pretty messy! It seems like there are only two regions I can use EQ on and that's 600 to 700 and 300 to 400. 600Hz to 700Hz show a more or less a linear response, but around 384Hz there is some resonances that we can EQ.
First I tried a PEQ of -2.5dB gain with Q = 2 to bring the response down.
View attachment 229504
Now you might expect that it would help with subjective sound quality but it barely did anything. We have to go back to point two that i shared, what we measure above 100Hz rarely show us what we actually hear. Above 100Hz we need to focus on Direct sound (speaker tonality) and very early reflections (less than 50ms late).
I applied Frequency dependent windowing (10 cycles) to the measurement without any PEQ at 384Hz and here is the result.
View attachment 229506
There is an early reflection that is causing 10dB peak at 384Hz! When all reflections arrive (measurement without FDW) it's only a couple of dB high because all the other late reflections are filling up the response else where.
While playing pink noise I decided that the sound was most neutral with a PEQ at 384Hz and a -7dB gain with Q of 2. This is what the regular steady state measurement looks like with that.
View attachment 229507
This is about how far you can go with regards to EQ. This is what 'Neutral' measures like in my room. There is nothing more to improve with regular IIR filters.
I'm thinking of dabbling with FIR to see if i can smooth the Excess Group delay between 100Hz and 200Hz but that's a project for another weekend. I cannot reiterate how dramatic pulling down early reflection peak was in terms of sound quality.