Thanks to
jaakkopasanen for all the incredible work on Impulcifer and making it all open-source.
I have just added a suggestion in GitHub:
Given that we likely know the approx frequency response of the headphones (thanks to the AutoEQ database), could we capture the frequency response of the mic capsules by holding the capsules directly to the headphones whilst playing the test sine through them?
Once we know the capsules' frequency response, we could use it for room correction without the hassle and cost of using a UMIK microphone.
That would almost never work and we wouldn't really know when it does. Your headphone and the one measured online are most likely different enough to make it audible(2dB is completely normal for most brands).
Then your ear isn't like the measurement rig. When you do a direct comparison between 2 headphones measured on the same system you still might have some variations from placement compared as to how you would wear it, but for the most part the variations are... I wouldn't say reliable, but usable at least. When you put the mic in your ear, you're adding the acoustic impact of your ear to the headphone measurement, and as capturing that is the main point of doing measurement with impulcifer, you wouldn't want to cancel the FR portion of it while trying to calibrate the mic. There is more to lose than to gain.
Also the graphs are supposedly made with the mic near the simulated eardrum while the mic you place in your ear remains at the entrance of the ear canal. So that's more stuff to try and compensate for.
IMO there are just too many variables. Even with the ability to measure my own headphone to get rid of the possible differences between my headphone and the online measurements, I have no idea how to objectively do what you're asking for. I would instead suggest one of 3 possible approaches:
1/ You wing it and EQ however you feel is improving the measurement(not great but if you like the result, it might be a progress).
2/ You hope that your equal loudness contour is average, and you use test tones to EQ based on when you can just barely notice the tones(you first set your amp as quietly as possible to just barely notice something and then EQ around that). Once it's done you "remove" the equal loudness curve(one for low level listening!!!) from your EQ and hopefully a bunch of things will be right that way.
3/ You use speakers that are kind of flat at your listening position to reference perceived flat.
We've mentioned Griesinger here a few times and his method is about calibrating in front of him(right in the middle with one speaker placed there). I think it's more effective for binaural tracks we would record ourselves at our ears. I'm not sure it's the right approach for us, but it's still probably better than nothing.
I think all approaches turn out to have one thing in common, you need to have some reference you can use even if it's just a mental one from musician training or whatever). The equal loudness curve seems like a good candidate as it might work alone(like in 1/), or be used for both the speaker sound and the headphone+impulcifer sound. If we can manage to match both to the equal loudness reference, then it means we have effectively linked both and we can generate an EQ for the differences.
I can't think of anything truly simple and fully automated.
@morgin I'm not ignoring you, I just didn't follow the amp market for years and I also don't know what your needs actually are. So I do what I'm best at in such a situation, I keep my mouth shut^_^.
