[markanini]

The bass rolloff might be because the bass just isn't there. You can do a huge correction to try and boost it, but it's going to be pretty distorted. If you want accurate sub bass, it's a lot better to use speakers rather than headphones. Sub bass isn't a strong point in headphones.

That's why I'm impressed by AKG K371. We will see if more headphones get released that compete in bass extension and low distortion.

 

[ctrlm]

This is interesting stuff, I'm definitely an EQ newbie as I don't use a computer for my music playback but I now have an RME ADI-2 which has 5 bands of parametric EQ + 2 shelf filters using the adjustable bass & tone controls. It's been fun to play around with the EQ on my Hifiman Anandas using all sorts of EQ recommendations but I soon started to realize that less is probably more and I think your theory of not applying EQ on higher frequencies is a sound one.

For my taste the Harman target with the obligatory low shelf boost is over the top and I prefer a flatter target up to 1k something like this with my supposed Ananda FR:

In other graphs there is a hump around the 200Hz mark but not here and my experience is that a 2dB cut there takes away a slight touch of muddiness. That dip at around 1.6 to 2k is even more pronounced in some FR graphs I've seen and the funny thing is that I have a hard time to hear any difference when I apply 3 to 4dB EQ there, maybe it needs more or maybe it really just doesn't matter? Or maybe I need to listen to the right instruments more intently

You talk about changes affecting following frequencies, I'm curious about that and wonder if for example....in theory if -2 to -3dB EQ was applied to the 3k area in the above, would that then accentuate the higher frequencies following it?

I see you use the same Q factor in your examples, is this due to only adjusting broad frequency issues?

Sorry for all the questions.....

 

[ctrlm]

I'll also add that I get confused by differences in FR graphs & EQ recommendations, yes I know that there are a lot of variables at play with positioning and measuring equipment. This example below of Oratory's EQ recs using (a) parametric EQ and (b) parametric EQ specifically for the RME ADI-2:

Quite a difference in some places for the same measured frequency response on the same headphone (mine) and especially the 6k & 6.4k adjustments in the 2nd pic. Apart from the fact that the 6k region isn't addressed at all in the first one, the extremely narrow band 6.4k adjustment must be questionable as that particular peak might actually be somewhere very different based on the variables mentioned above?

 

[markanini]

You talk about changes affecting following frequencies, I'm curious about that and wonder if for example....in theory if -2 to -3dB EQ was applied to the 3k area in the above, would that then accentuate the higher frequencies following it?

I see you use the same Q factor in your examples, is this due to only adjusting broad frequency issues?

3kHz is a frequency where the ear is very sensitive to change, cutting it will accentuate all frequencies around it. It would also sound similar to boost at 250Hz, try both and see which sounds best to you.

I use constant Q because its accurate enough for non-individual measurements and facilitates quick comparisons. My guide deals with the law of diminishing returns.

 

[ADUHF]

Howdy again markanini.

I have used and tried a number of different approaches to EQ-ing. And continue to experiment with new ideas along these lines. I'm not sure that I've used an iterative approach exactly like what you describe here. I have used an approach where I listen to and EQ a number of different songs though. And then compute an average EQ curve for my headphones based on those. The table posted here was one example of this...

https://www.head-fi.org/threads/bey...pressions-thread.702948/page-31#post-16046298

What I've been trying to do lately though is figure out an easy way of combining the sound signatures of multiple headphones together. Perhaps using kind of an iterative approach somewhat akin to what you describe here. I believe there is an an app that can already do something like this. Which allows you to EQ your headphones to sound like other headphones. And also to combine the responses of several different headphones together, which is pretty cool.

The simplest way to do something like this would probably be with a spreadsheet. I think I've figured out a temporary work around that I can use in Equalizer APO's Configuration Editor though, using it's stacking feature, to potentially save me some of the rather heavy-duty number crunching that's involved with this. (I may try this approach for computing the average sound power for different sets of loudspeakers as well!)

If I stack all the EQ or headphones response curves together in the same project, the Configuration Editor's Analysis Panel will basically show me the combined result of them all added together. At that point it should be fairly simple to compute the average by simply dividing (or scaling) the final result by the number of EQ or headphone graphs in the stack. There are a couple other steps involved. But that's the basic idea.

This should hopefully allow me to try more combinations of EQ, or response curves without having to manually add up the values for all the points, as I've been doin up to now. Which is one of the things which has been slowing down alot of my graphing lately.

 

[bigshot]

It's better to calibrate to a single set of headphones. If you average a bunch of different curves, you'll come up with an average curve, not a precise one. That's why the Harman curve is just a starting point. It's an average, not specific curve. You calibrate to Harman, then fine tune by ear to accommodate the differences between the coloration of your cans and the variance of your particular ear canal configuration.

 

[ADUHF]

It's better to calibrate to a single set of headphones. If you average a bunch of different curves, you'll come up with an average curve, not a precise one.

I hear what you are saying, bigshot.

The approach that I'll be using for this will be a little more involved than what I described above though. And may involve creating a number of different EQ curves for my Beyer DT-770's, that would match the responses of various different headphones. Or various different groups of headphones, with slightly different (but still fairly neutral-ish) sound signatures. And then sort of mixing and matching together the ones which sound the best with a wide variety of music.

I am still in the very early stages of trying to figure out the best way to go about this though.

Calculating the average of multiple response curves is one of the most time-consuming and labor-intensives parts of my current process. So I'm hoping that the above work-around will maybe speed that process up a bit. I'll try a few experiments though, and see how it goes.

 

[bigshot]

Ah. So you aren't looking for a target curve to suit the music. You're looking to emulate various headphone curves. I do it the other way around. I look for headphones that match the curve that suits the music. Balanced isn't a range. It's a specific calibration. Once I get there, I'm done and I can listen to music and not worry about EQ any more.

 

[ADUHF]

Ah. So you aren't looking for a target curve to suit the music. You're looking to emulate various headphone curves. I do it the other way around. I look for headphones that match the curve that suits the music. Balanced isn't a range. It's a specific calibration. Once I get there, I'm done and I can listen to music and not worry about EQ any more.

I think you've got the general idea. I want to try a variety of different sound signatures with my Beyer DT-770's, to see which ones I like the best. And maybe that will help me to narrow down what "my sound" is a little better. Or at least what my sound is with the DT-770's anyway.

Whether it'll also translate to other headphones though, I don't know. (?)

I also just need a faster way of combining multiple curves together than what I've been doing up to now.

 

[bigshot]

For me, I get confused listening to a bunch of different sound signatures. I find my personal target by calibrating to an average and then making small modifications and living with it for a while. Then I make another small modification and let it soak in. Lots of little baby steps leading towards a target. If I do it in chunks or apply random curves, I just wander and never get anywhere.

 

[markanini]

The main issue is Beyerdynamics unit variation is known to be pretty high. It could work with individual measurements.

 

[ADUHF]

For me, I get confused listening to a bunch of different sound signatures. I find my personal target by calibrating to an average and then making small modifications and living with it for a while. Then I make another small modification and let it soak in. Lots of little baby steps leading towards a target. If I do it in chunks or apply random curves, I just wander and never get anywhere.

I have used incremental approaches like this as well. Since I only own two HPs (and rarely ever use my M50x), I'm sort of curious to try out some different sound sigs though with these HPs. Should be fun, and interesting. And hopefully also educational.

The curves won't be random btw. They will all be fairly neutral. Some may be slightly more U-shaped though. Or more rolled off in the bass. Or brighter in the treble, or upper mids, and so forth. That's sort of the plan anyway. Whether or not it'll work, I don't really know yet.

Since I've already done a fair number of EQ tests with these HP's, I'd expect to narrow down which sound sigs work well, and which don't fairly quickly. Maybe I'll be surprised though, and discover something new and a bit different that I haven't heard before.

 

[bigshot]

Most mid range headphones are going to have sample deviation from 3 to 5dB from each other at some point in the range. The difference between high end and mid range headphones isn't sound quality, it's consistency.

 

[bigshot]

The curves won't be random btw. They will all be fairly neutral.

Neutrality isn't a range. It's a specific target. "Fairly neutral" means "sort of accurate".

 

[ADUHF]

The main issue is Beyerdynamics unit variation is known to be pretty high. It could work with individual measurements.

Most mid range headphones are going to have sample deviation from 3 to 5dB from each other at some point in the range. The difference between high end and mid range headphones isn't sound quality, it's consistency.

There may be a little sample/unit variation on the Beyers. I think some of the variation that you see on some Beyer plots (like the ones below) may be due more to the condition/wear of the velour pads though.