Hello,
Some months ago I wanted to equalize my sennheiser hd457 headphones. After some thinking I did something really crazy.
Here is what I used:
A cheapo Grundig microphone that came with some portable tape recorder
An equalizer that did spectral matching
A Pink noise sample
A Deconvolver

This may have been done before (probably it was), but I tried to look for such articles and didn't find any, so I'm posting this.

First I found a site containing headphone response curves. I don't remember the name of the site at the moment. I think they sell headphones and their own preamps. Anyway I replicated the curve in an equalizer as accurately as I could, but I couldn't do it perfectly.

I took the microphone and blew into it, captured the response I got and matched it against the pink noise curve, so that the response of the microphone would look like the pink noise curve. I saved it as it would be used later.

Then, I put the mic about 1 or 2 cm from one of the headphones and run a deconvolver.

What I did next was using a spectral matching eq to make the file generated by the deconvolver "half as strong". What I mean by this is that the eq should use the file from the convolver, but apply only half as much equalization. Than I used the correction file for the microphone. I saved the curve and applied it to the eq curve I made using the graphs from the site I mentioned. I got an almost straight line !!! Well... almost. There were a couple of peaks that I couldn't reproduce using just the graphs, as the graphs were very small and I couldn't say which frequency it was, but they the peaks were on the graph!!! There was just a bit of inconsistency on the low end (sub 40hz) and top end (more than 12khz) where this failed, because the suggested values were absurd, so I equed it a bit to taste.

I think I used 2 eq instances. One with my initial hand drawn curve and the other with the "correction" curve. I generated an impuse response file that I'm using to this day.

Again, this may have been done in the past, if so, sorry, didn't find such articles.

Comments are welcome.