I know this an old thread, but I just ran across it and thought I'd share an anecdote about the topic. My father has very bad hearing loss in both ears and has been wearing hearing aids for the last ~4 years. He says music sounds awful through his hearing aids, so I suggested I could try to compensate using a parametric eq in JRiver. I took his audiogram from the doctor and created several different inverse peq curves to compensate. Varying from complete correction to mild boosts in the higher frequencies. He did an informal A/B test with each filter vs flat, while listening through headphones without his hearing aids. My initial hypothesis was that one of the milder correction curves would be his preferred (the full correction curve sounded hideous to my ears). Interestingly, he preferred the flat response in every instance. The possibilities why a multiple--it could be that the eq wasn't precise enough, or that his hearing loss can't be compensated for in frequency boosts alone, etc. The simplest theory, however, is that he's had bad hearing loss for so long (decades) that he's just more comfortable with the depressed highs in the flat response. As long as he listens through headphones or IEMs that isolated him from outside noise and the complexities of room acoustics, he can still hear enough of the music to enjoy it.