Attempt at measuring personalized headphone response curve

Last thread I posted was about trying to tweak a pair of headphones using in-ear mics. I used the Harman response curve as my goal and equalized to get as close as I could. I had wanted to go to a high end audio shop and see what kind of measurements I could get from some of the speakers there, but my email request was ignored by the local shop, so I decided to use my own speakers.

As others I think have found, consistent in-ear measurements can be tricky. Here are my results:


image hosting 15mb

The dotted line is a 6 period moving average of the average of all of the in ear results.

My process was to set my mic on axis with the tweeter on one of my monitors, and compare it to the mic in my ear at the same spot.

Getting consistent measurements from the tweeter was pretty easy, I did a few tests where I measured, took the mic from it's spot, put it back as close as I could and measured again. The results were very close.

Trying to hold my head at the same spot and the right angle was not the most scientific process. I made an attempt to move my head so the sound hit my ear at different angles, so that accounts for some variance in the results. The last two tries were set with my face turned more towards the speaker.

Despite the inconsistency of the results, I did see a pattern that somewhat matches the headphones response curves. It also reflects a null at about 7500 or so, which I always see in my in-ear measurements that doesn't show up on the generic response curve.

Another day I'll try to find a better way of setting my head at the angle that the hd800 hits my ears and see if the results come out better.

 

Took another shot at this. This time I tested farther away in hopes of reducing an reflections off walls or objects. Took measurements in groups. spkdiff is the difference between taking two consecutive measurements of my speaker with the mic by itself. Measurement 1 I was turned towards the speaker quite a bit. The others are pairs of measurements (2&3, 4&5, 6&7), between which I'd take the mic out of my ear. You can see how similar they are when I'm at the same place in front of the speaker. The variation between the pairs is due to microphone variability (which I tried to measure in spkdif) and changing the angle of my head between measurements.








Between 6 and 7, I readjusted the mic, so that is to blame for the differences here being greater.

The biggest variable looks to be how the mic is placed in my ear. If I can get an in ear mold of some sort, I expect I can get consecutive measurements very close to each other, and then hopefully try to get a personalized in ear map within a few db across the range.

 

Please do not EQ your headphone to match the FR of your speakers. That won't sound good. There's some special psychoacoustic stuff involved that makes your brain realize it's listening to speakers - simply matching the FR won't do the same thing for headphones. Here's an article on IF: http://www.innerfidelity.com/content/aes-headphone-technology-conference-highlight-paper
(This is also why OOYH sounds weird. Unless you are Darin your brain will know that it's listening to headphones and the tonality will be all messed up.)
Essentially this says that you need to get rid of the bump around 4kHz again. You'll also see that when you measure the same speakers in different rooms the FR will change much more than with omni mic measurements. This is a good thing because they actually do sound that different.

About the mic position ... I wouldn't worry too much. Try to find something that you like and then stick to it.

As for the speakers ... as long as you find your speakers reasonably real sounding tonally then you're probably on the right track. Simply going into a store and measuring 100k$ speakers will probably not get you closer to a good tonal balance than cheap speakers with careful setup. Actually a lot of the expensive speakers tend to be really weird sounding.

In terms of what angle gives the best tonality - it's really up to you. I like to use the "nose facing speaker" as a guideline for the upper midrange

Otherwise, looks like your ear is quite different from mine. I always get peaks at 5, 8 and 10kHz which you don't seem to have. Interesting.