Sure... some of the other steps below is just to improve readability of the measurements:
1) Apply some smoothing to make the magnitude and phase readable
View attachment 353420 View attachment 353421
2) If the phases aren't aligned, apply 'cross corr align' before
vector averaging the left and right measurements -- I just don't want to look at each channel individually for the sake of expediency!
View attachment 353422 View attachment 353423 View attachment 353424
3) If the phase and IR timing still looks
off, apply 'estimate IR delay' and/or 'invert polarity'
View attachment 353427 View attachment 353425 View attachment 353426
4) Let REW calculate or create a 'minimum phase version'
copy of the LR vector averaged measurement
View attachment 353430
5) Apply
frequency dependent windowing (the 'IR Windows' button right beside 'SPL meter') to both original and MP version copy
View attachment 353432 View attachment 353431
Dips or reduction in level occurs when delayed pressure/energy outside the specified window is removed -- usually caused by reflections. The MP version is what the transfer function would be if the (peak) "energy" along the time domain axis were theoretically pulled in or compressed much closer t=0 i.e. as if the measurement were converted to its
minimum phase equivalent and
excess phase were removed
. The larger the dip, the greater the delay in that frequency range and/or 'smearing' of sound across time.
View attachment 353433 View attachment 353434View attachment 353438 View attachment 353440
*
In general, when using
FDW for EQ purposes -- in order to avoid over-correcting or under-correcting -- more cycles (e.g. 15) should be applied to the lower half of the response while fewer cycles should be used when approaching the highest frequencies.
**Oh, yeah, you're supposed to avoid filling-in such dips as that will only increase/worsen the problematic non-minimum phase or the reflected-delayed energy!