However it seems at the same time to compress the dynamics as well. And especially with recordings with a large dynamic range, such as classical music, this sounds just awful.
I'm also not certain that it's just a FR change. I haven't bothered devising some tests to that effect since I don't really like the results with headphones accommodations on anyway, but perhaps measuring signals (which ones ?) at various levels and comparing the delta with headphones accommodations on vs off could reveal additional behaviours we're not aware of ATM ?