The Harman target lives of a debunked myth that headphones/IEM's need 6db extra bass to mimic the output of a Speaker. But like you said 90% of every Speaker have a -12.8db roll off under 65Hz which something headphones don't have a issue with like the ER4SR/DT880. So in a way the ER4SR & DT880 will sound like a Speaker with a woofer, The HT is just that a preference tuning I find the 3 ~ 4db boost on the ER4XR more than enough assuming I cut the 190Hz(Q 1.6) area by -1.9db.So it's a preference of a preference which tries to mimic something which it isn't and changed true time. Not very convincing, I hope you would agree. Only thing I agree is that I would try to boost speakers a bit at 30 Hz up to 60~70 in falling down manner and only if driver can take it. Wouldn't tame it down before 8 KHz and only a little if they feel bright. I even prefer slight boots in mids. Now the hedaphones are entirely different thing superior when it comes to lows reproduction (because energy there fades fast and distance makes a big difference). So why would I want them to mimic speakers (especially if they can get all the way down to 20 Hz)? This is a big question for both parties (Harman and Fostex - Foster alike) all do they "prefer" very different "preferred" curve they try to do the sama with hedaphones. For me old raw diffuse field (study dating back to 80's) have much more sense preference aside and letting drivers do what they can.
I am actualy taking about below 200Hz content in the male voice. it just sounds so unatural. and spoken human is the best test you can have, since it is the sound we most know how it should sound.
I've noticed that too doing a 190Hz cut of -2db helps or do a 55 ~ 79Hz Low shelf since sub bass doesn't get in the way like mid to upper bass does.