Personalized Equalization with in-ear mics

Discussion in 'Computer Audiophile: Software, Configs, Tools' started by randomg, Nov 16, 2016.

  1. randomg

    randomg New

    Banned
    Joined:
    Nov 3, 2015
    Likes Received:
    4
    Trophy Points:
    3
    Edit: turns out I'm too wordy. TL;DR:

    The HD800 has 4 places where the FR isn't perfect: the sub-bass, the dip at 1-2 Khz. the peaks at 6-7 Khz and 9-11 Khz. Using in ear microphones and an equalizer and Tyll's measurements as a goal, I can fix how these spots behave in my own ears with some sort of objectivity and confidence beyond what I'd normally get from listening to sine sweeps and trying to hear where it doesn't sound flat. The same process should work similarly for any over the ear headphones. Below are pictures of the results.

    Original Post:

    I think the topic is academic sounding enough. I've thought a lot about equalization the past few years, in particular in regards to my pair of HD800. It's so close to being as good as you can get, and yet so far. As far as I can tell, besides its FR, it bests or ties anything else out there when it comes to objective measurements.

    I'm using DMG Audio Equilibrium. I could have experimented more with other equalizers, but when coming from Electri-Q, the difference was so noticeable I just decided I'd go with popular opinion and buy what people consider the best one. For reference, I'm using the Prodigy HD2 Advance as my DAC and Amp.

    I decided to take a $90 risk and purchase a pair of in ear microphones, with the goal of matching MY HD800s to MY ears. I've tried to equalize based on my own ears in the past, and was always unhappy with the results. I did that tuning with Electri-Q, so maybe doing it again with Equilibrium might go much better. But one of the hardest parts was knowing when to stop. You can listen and tweak forever if you let yourself, and with listening fatigue as a variable, it was just too imprecise for me to ever be happy with it.

    The process I came up with is, play a sine sweep of part of the frequency range, record the result with the microphones in my ears and the headphones on, use Equilibrium to try to improve the result, and then repeat.

    Here is a list of requirements I came up with for this experiment to be successful, and then an explanation of why I think the requirements were met:

    1) Repeated measurements without changing variables must show the same results
    If I couldn't press record and listen to the same thing again and get the same result, this wasn't going to work. Thankfully if I kept the room quiet and myself from making quick movements to move the mics or headphones, this was not an issue, measurements were repeatable*. One other thing to note, when measuring my roommate, I got a repeatably different response from her ears than from mine.

    2) The ideal frequency response is known
    This one's tough. When I first started measuring, I was hoping for something that was close to what Innerfidelity measured for the HD800s. That is not what I had. I considered the idea of going to some super high end audio shop and getting some base measurements of what the ideal frequency response is when my ears are listening with these mics in them, but decided it'd be tough to get a true ideal from one of those places. I settled on treating Innerfidelity as accurate, meaning that what I was really trying to do was get rid of that 1-2 dip, 5-7 peak, and 9-11 peaks (the 100 Hz and below dip I just copied Tyll's HD800S article on). Any differences between what I measured and what Tyll's expensive head measured, I treated as a mapping of what I was measuring to what Tyll measured. The map ended up being that these mics measured a slow rise from 1500-5000 instead of a 3-4 peak, but otherwise were just about the same as the expensive head.

    So my goal became to remove 6 db off the peaks between 5-7 and 9-11 and fill in the very small dip between 800-2000. Equilibrium also let me tune left and right separately, and the differences between sides were pretty significant in the treble, so that came in pretty handy. Also, I went with what Tyll measured for his head acoustics instead of the adjustment he normally applies to his graphs. The differences are minor, but the hd800 is much closer to what he measured for the 1500-3000 rise in Harman's room than what Tyll's adjusted HD800 FR graphs say.

    3) A flat sine sweep doesn't come at the cost of other sound quality aspects
    I'm letting the HD800s and Equilibrium handle this one, my verdict is that they did just fine.


    *I ended up having to cut off the rubber part that holds them in my ear so that I could actually put the mic in my ear canal. This helped a lot with getting repeatable measurements.

    Anyway, I have a lot more to say about what program I used, my techniques, etc. But I'll just post my results and see if anyone is interested in trying this themselves before I spend more time. The following are sweeps from 1-15K:

    [​IMG]
    Left Before
    [​IMG]
    Left After
    [​IMG]
    Right Before
    [​IMG]
    Right After

    Anyway, despite this costing me ~$310 and 5 hours (now that I know the process, I could do it in less than 2 hours), I'd wager it's better than an HD800 without EQ on any amp or with any mod. Using another EQ could potentially do just as well, but for me, knowing I have the best one makes me feel better about the result....

    The following is my subjective take on the results. I haven't had the feelings I've had in the past when EQing or modding, which are either "Ow, that 6K peak still hurts on this song" or "I'm missing detail I used to hear before I tried to cut back the treble". Not sure I can say this with confidence, but I've had the feeling that I hear things in the music I haven't heard before, I imagine because they were drowned out by the treble peaks before.

    Edit: I'm having second thoughts about this last paragraph, I don't want people spending money on in ear mics only to be disappointed with the results, but for me, $90-$300 to add to the over a grand already spent on the headphones was worth it to me.
     
    Last edited: Nov 19, 2016
  2. randomg

    randomg New

    Banned
    Joined:
    Nov 3, 2015
    Likes Received:
    4
    Trophy Points:
    3
    I'm a little surprised I didn't get any replies. I figured I'd get something telling me either it's a good idea, it's a bad idea, or it's been done before. Maybe this should have gone in the headphone Forum?
     
  3. shipsupt

    shipsupt Admin

    Staff Member Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    1,119
    Trophy Points:
    93
    Location:
    Bay Area CA
    No, not the headphone forum. This isn't a bad place for it, or maybe the general forum. Give it some time, people are likely digesting what you wrote.
     
  4. TwoEars

    TwoEars Friend

    Pyrate
    Joined:
    Feb 2, 2016
    Likes Received:
    2,006
    Trophy Points:
    113
    @randomg

    People have very limited attention spans these days. You need to get better at distilling what you want to say.

    Your headline tells me nothing about what your intention or purpose is. Then the first three paragrahaps tells me nothing about what your intention or purpose is. By now you've lost 99.8% of all readers.

    Your headline should give a clear understanding of what you're trying to achieve - if it's witty it's a bonus. If it's too complicated to be a headline then make the headline more general. Then your first paragraph must contain some kind of plan, intention or mission statement. Don't be vague and don't beat about the bush.
     
  5. Thad E Ginathom

    Thad E Ginathom Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    14,134
    Trophy Points:
    113
    Location:
    India
    Well, I liked it. Thank you @randomg!

    I could have even coped with maybe 20% more practical detail. My interest is capped as all is not well inside the ear drum.

    Of course, many people have imperfect hearing. Just as many people with speakers have imperfect rooms too. But all this stuff has a space and a use, and is, according to my humble, no-longer-audiophile, opinion, a lot more useful than much of the gear nervosa.
     
  6. randomg

    randomg New

    Banned
    Joined:
    Nov 3, 2015
    Likes Received:
    4
    Trophy Points:
    3
    Thanks for the comments, I threw in a TL;DR at the beginning for people looking to skim.
     
  7. Hekeli

    Hekeli Facebook Friend

    Contributor
    Joined:
    Sep 28, 2015
    Likes Received:
    149
    Trophy Points:
    43
    Location:
    Finland
    Do you have any graphs about the difference? Other than those unreadable Audacity ones, you should use atleast REW for measurements..
     
  8. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    1) Don't OCD.

    2) I wouldn't equalize full-size open headphones with in-ear microphones TBH.

    3) I would experiment with free space measurements for anything above 2 or 3 kHz, and then baffle measurements for anything below that.

    4) You could approximate things using a baffle with some high frequency sound absorption w/o seal loss. But you might not get 100% there.

    5) You might want to target the B&K curve. @Marvey kind of preached that for a while. Well, I tried equalizing flat, and indeed ended up liking B&K better.

    6) Read 1)

    BTW, IMO the HD800 are not best or tie anything else out there when it comes to objective measurements. They maybe fixable through equalization. But they do have significant distortion in the low frequencies when driven hard. "Objectively" they suck. I've got no idea why they cost what they cost. If all else fails sell them.

    YMMV, others thing they are awesomest among the awesome, or so.
     
  9. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    You could've most likely gotten better results with one of the 1$ WM-61A mics inserted into an eartip. As UB said, use REW. He doesn't seem to like the in-ear measurements, but personally I think they're usually the closest to what I hear uncalibrated. EQing for a flat FR at the ear-canal opening will most likely give you a better tonal balance than using Tyll's plots (his compensation curve isn't the best, so it needs some guessing anyway) and will almost certainly give you a much much better tonal balance than using coupler measurements without compensation. As you've noticed, there is no 10db 4kHz dip with the stock HD800. The B&K target on a coupler works relatively well for the HD600, but doesn't work at all with the HD800.

    As far as couplers, I feel a better, more universal coupler used by the whole SBAF measuring crew would be a good step forward, but it's really hard to make one universal coupler that everyone could agree on. Too many tradeoffs.


    @Hands made a detailed list of things to look out for with in-ear microphone measurements, but I can't seem to find it right now.
     
  10. Kattefjaes

    Kattefjaes Mostly Harmless

    Pyrate
    Joined:
    Sep 5, 2016
    Likes Received:
    4,521
    Trophy Points:
    113
    Location:
    London, UK
    It would be nice, though- especially for those of us who wouldn't trust a home-made one, necessarily. It'd be useful to have a trustworthy and affordable coupler, even just for headphone modding fun.
     
    Last edited: Nov 20, 2016
  11. randomg

    randomg New

    Banned
    Joined:
    Nov 3, 2015
    Likes Received:
    4
    Trophy Points:
    3
    I'll try to get some of those this week. I'll check out REW.

    Why not?

    Also, maybe I'm not reading the right measurements, I thought I had a general idea from following head-fi, changestar, innerfidelity and here. Is there some other place that has measurements showing the HD800 not being world class, besides loud bass and a few spots in the FR?

    Probably should have done more research before jumping in I guess. I missed that in ear EQ was a thing people did.
     
  12. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Regarding using an ear? You can do in-ear measurements if you like. Then you would have to compensate to account for the gain due to the outter ear shape. Many folks use different kinds of compensation, except the one that best represents the outter ear they used. Which seems a bit counter intuitive to me.

    If you do compensate by removing the gain obtained by the outter ear actually used in the measurement, then why even use the ear measurement in the first place?

    What to me remains a bit more problematic is the interactions between the driver and the volume that it has to drive.

    I prefer not to introduce the gain of the outter ear by using an ear, and remove it later through compensation. But you might feel different about it.

    Don't get me started about measurement repeat-ability when using an ear.

    Could be there are interactions hard to remove when using ANY coupler. Which is why I'm thinking about free air measurements for upper frequencies.

    What in the measurements suggests the HD800 are better than it's peers? Distortion? Nope - It's not horrid though. Frequency Response? Nope - Too analytic. Bass extension? Nope - Most open cans have a hard time here. Balance? Nope - Again, too analytic. Impulse Response? Nope.

    What do you feel the HD800 can do that other cans struggle?
     
    Last edited: Nov 20, 2016
  13. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    You can jump in any time you like BTW. Try what works for you.

    As far as in ear EQ, some people do it. Some people don't.
     
  14. TwoEars

    TwoEars Friend

    Pyrate
    Joined:
    Feb 2, 2016
    Likes Received:
    2,006
    Trophy Points:
    113
    For me the concept of in-ear EQ is somewhat like quantum mechanics. By the time you've managed to measure it, you've already changed it.

    And I've stopped believing in EQ'ing for something like "neutral" or an "absolute truth" myself, I don't think such a thing exists. I'd say - if you're going to use EQ Just dial it in the way you like and enjoy the music.

    Not to say that trying to find "the perfect EQ" can't be interesting from a scientific or experimental perspective, I just think that in the end it has very little to do with actually enjoying music. And you might find you prefer something else than what is considered "neutral" anyway.
     
  15. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    I think we can achieve it. Or at least get close to it. I just haven't got around it. Maybe later today, or tomorrow, or who knows :)
     
  16. randomg

    randomg New

    Banned
    Joined:
    Nov 3, 2015
    Likes Received:
    4
    Trophy Points:
    3
    I found the problem areas move around dependent on the ear, so I really used it as a way of finding exactly where those areas were (where exactly that 6Khz peak was), and then as a way of being able to actually check that the given peak is gone after I'm done. How would I know where that peak was if I wasn't measuring my own ear?

    I guess I'll look around,

    If I compare innerfidelity hd800 vs sr009 vs he1000, distortion is close between the sr009 and hd800, definitely goes to the sr009 at high volume, but overall is lower for the hd800 at 90. The impulse response isn't as quick on the hd800, but it's plenty quick to catch any rise below 20khz, and returns back to 0 better than both of the others. It follows the 300 hz square wave better than the other two also. As far as I've seen it's csd, especially when dropped down below the typical 36 db, is better than any other headphone. Bass extension, isn't that frequency response? I'm not sure what you mean by balance, matching between drivers, or are you talking about another FR issue?
     
    Last edited: Nov 20, 2016
  17. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    I think one should focus on trying to equalize/correct issues with the headphone independent of ones ears.

    Consider speakers. Depending on ones head orientation, a flat quality speaker will sound different. If one faces it, it will sound like a flat speaker in front, if one looks away from it, it will sound like a flat speaker on the side.

    Things will sound different depending on orientation, and perhaps ear shape in the absolute sense. But I don't think one should be equalizing ears. One should equalize headphones, or speakers.

    The SR009 destroys the HD800 in distortion. Is not even funny. The HE1000 is a joke. It is a horrid distortion piece of shit relative to both the HD800 and the SR009.

    Distortion is a function of volume. At low volume distortion is not an issue with most cans.

    The impulse response is deceiving. It depends on the coupler. That said the HD800 is definitively bandlimited (like all other cans), and it shows.

    The square wave at 300 Hz tells me nothing the frequency response doesn't already tell me.

    The CSD seems relatively clean. But it may not be as immediate and clean in the 8 kHz range. I do not see anything stellar. Just not fucked up.

    Bass extension rolls down starting at 100 Hz, and yes, that shows up in the frequency response.

    By balance I mean the balance of the frequency response. It's tilted up.

    The drivers tend to be well matched.

    If you like your HD800s life is good. But I don't consider their measurements impressive.
     
    Last edited: Nov 20, 2016
  18. Hekeli

    Hekeli Facebook Friend

    Contributor
    Joined:
    Sep 28, 2015
    Likes Received:
    149
    Trophy Points:
    43
    Location:
    Finland
    Not sure what the point here is. Isn't the goal of equalizing to fix how things sound to your hearing? Ears are part of the equation. If it helps getting to the goal using in-ear mics or whatever, that's good yes? What's the point of an academic goal of both transducers sounding flat or whetever curve, if the result doesn't sound good? Perhaps one has hearing damage in other ear etc. Of no ear lobe at all.

    I've noticed that when listening to speakers I have to look right about 5-10 degrees for things to sound balanced. Pretty annoying. Too bad I don't have my Realiser mics anymore, would be fun to test how it looks when measuring.
     
  19. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Perhaps I'm a bit dense and missing your point. That said:

    If you want your audio reproduction system to sound like real life sounds sound to you, then you may equalize accordingly.

    If you want to compensate for other issues, then you equalize accordingly.

    Hearing issues apply to your audio reproduction system and every day life. Your brain should compensate as much as possible for every day life. If that is not enough, a hearing aid may be needed. I don't see the point of taking it off when you are listening to your stereo.

    Going about the Realiser IMO is pure speculation, because I don't know exactly what they do, and unless you are one of the designers I doubt you know either. My best educated guess is that they look at deltas at different positions and apply a coarse spatial set of equalizers accordingly. Whatever they do, it works. But I doubt they are using absolute measurements. If they do, then IMO what makes it work is the relative aspect of the measurements at different angles.
     
  20. Hekeli

    Hekeli Facebook Friend

    Contributor
    Joined:
    Sep 28, 2015
    Likes Received:
    149
    Trophy Points:
    43
    Location:
    Finland
    Brains should probably compensate for bad DACs then too. Why are we on this forum?

    Pretty handily you dismissed my looking angle too. Perhaps my ear lobes make things sound a bit out of balance. I should just accept that "it happens"?

    Sorry I was referring to the handy calibrated in-ear mics they provide. I've made many headphone measurements with them.
     

Share This Page