Headphones Frequency Response: Challenges & Solutions

Discussion in 'Measurement Techniques Discussion' started by samvafaei, Jun 19, 2017.

  1. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    Hi SBAF!

    We have made a video explaining some of the common issues in making frequency response measurements, and the solutions that we have come up with. These include bass leakage/coupling issues, treble inconsistencies, and target response curves.

    Do you have any comments on how we could improve our measurements of headphones even further? Some of the stuff I explain with video are a bit simplified to keep the video from getting too long. But we can go in more detail here if you guys are interested.



    Sam
     
  2. numbercube

    numbercube Acquaintance

    Joined:
    Sep 29, 2015
    Likes Received:
    98
    Trophy Points:
    18
    Location:
    Germany
    Last edited: Jun 19, 2017
  3. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    Thanks for your reply. I'm interested in your opinion as to why the 1974 B&K target is preferred to Harman's in-room target here? Is that your personal preference too, or just the majority consensus at SBAF?

    I'm aware of the thread that you linked to, but our video doesn't really cover the same subject. It's just a coincidence that they came out around the same time.
     
  4. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,589
    Trophy Points:
    113
    Location:
    near Munich, Germany
    That was a great video and explained very well some of the issues with measuring headphones. I'm still baffled as to why your blocked ear canal measurements look like that and not like the results me or @Hands get. Maybe you should experiment with less insertion depth, I'm not sure.

    Most people here (me included) will find the Harman target too bassy. The HD600 and HD650 both have a mid-bass hump that the target curve takes away. I feel a bassy response should be a personal choice and not part of a target response for the measurements. I would leave the area below 500Hz flat in the compensation. With the bass-hump in the target response I'll have to completely ignore your results below about 400Hz. Maybe you could show the plots with and without the Harman target bass emphasis.

    For now looking at mainly your HD800S, HD600 and HD650 measurements to compare with mine and others:
    Otherwise I find your measurements to show a bit too much in the 2kHz region. I find the HD600 slightly recessed around 2kHz, but not everyone will agree. Phones like the Elear and the K701 have way too much 2kHz. This also depends on the pads. The region above 4kHz looks a bit recessed compared to what I'd find neutral, or in other words the target seems slightly bright to me. 1 or 2db go a long way here. But generally the results are a bit weird in that the HD800S shows a flat line from 1-4kHz. I would still raise the region, especially around 4-5kHz a bit, even though that might not make every measurement more accurate. In this case I feel I'd actually agree with your in-ear measurements, at least with compensation, more than your dummy head measurements.
    The difference in the upper midrange also seems to be one of the major differences between your and the harman target curve. Your target curve seems to be consistently lower in the 2-5kHz region and merging the two target curves at 3-5kHz might get you slightly better results. I previously also thought the Harman target had too much 1kHz for my tastes, at least based on the measurements Tyll showed us here. For example these show a distinctly different shape. The 1kHz region might still look a tiny bit low in your measurements. In general terms I'd prefer slightly less 1-1.5kHz, slightly more 2kHz and less 3-4/5kHz than the HD600 for a target curve. And I find the HD600 a bit bright and the HD650 way too dark for my tastes. I also find both to have too much mid-bass.

    Other than the 2kHz region looking maybe 2db high to me and the bass hump in the target response that I feel is absolutely not needed your results look pretty good to me. Would be interesting to see what the others think.


    EDIT: Looking at these for example: https://clarityfidelity.blogspot.de/2016/04/sennheiser-hd-600-over-ear-headphones.html
    I actually seem to prefer the old Olive Welti target or something in-between, but of course completely flat below 500Hz. The Harman target seems to be brighter, but I kind of prefer the look of the Olive Welti target between 1-4kHz. Above 5kHz Harman might be better.

    One thing that's important to consider is that music is mixed on speakers. With speakers the tonality changes depending on how sounds are panned (for example look at this graph in the 2nd IF article I linked). This I think is compensated in the mastering and makes the FR on headphones off on stuff that is panned to one side, or at least that's what I hear*. I feel like stuff that is panned to one side sounds tinnier than it would on speakers, even if the center stage sounds similar. Slightly less 2-4kHz can sometimes compensate for this.

    *It's probably the lack of crossfeed.
     
    Last edited: Jun 20, 2017
  5. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    Thanks for your thorough reply Serious :)

    I haven't seen the blocked ear measurements that you mentioned, but I'm interested to see and compare them to ours. We have calibrated our microphones to match the response of the ear simulator in the HMS and they work basically flawlessly. The bass of headphones measured on the HMS with and without the ear mic look the same now. The position of the microphone is actually a non-issue, since we are dealing with Bass frequencies and they have very long wavelengths. You can actually flip the microphone, or even let it dangle inside the ear cup, and wouldn't make any difference in the Bass reading!

    The first thing to consider about the Harman curve, is that it was based on measurements done with a GRAS 45CA, which is quite different from the HMS. Below you can see the difference between the measurement of the HD 600 on our rig, vs Harman's rig (check out the 10KHz reversal!). We exchanged a few more headphones measurements than just the HD 600, and none of the responses were even close. So manufacturing tolerance is most likely not an issue (and if you compare my HD 600 to Tyll's you won't find much difference)

    [​IMG]

    And the difference above is only in the Outer Ear region measured by headphones, and not in the full HRTF. Our head has shoulders, but the 45CA is just a cylinder, so if you measure them with speakers at a distance, the difference will be even larger. I think in the end the Headphones Target Response will be more of a "standard" than a fixed curve, since each head and torso simulator (HATS) should have its own target curve.

    My personal opinion is that individual preferences can't and shouldn't be more than a couple of dBs. At least that's the extent of variance in the preference of audio engineers, which I believe should be the subjects for these kinds of tests. I don't think untrained listeners are good subjects for coming up with a target curve (that's why I also prefer the old 2013 Harman curve to the 2015 curve).

    I don't agree that the bass bump in the Harman curve (and ours) should be optional. Neither the science behind our headphones target curves, nor our listening tests support such an idea. Even our untrained listener's that we used for verifying our target curve preferred the bass bump in the target, and disliked the bump around 200Hz that most Sennheisers have. Part of the issue here is that the tracks people use as reference are not mixed all in the same room. My personal opinion is that for a proper control test, the reference tracks should be mixed in the same room that the target curve was derived from, and then be signed off on by a number of engineers in terms of bass/treble balance.

    Regarding the issues you mentioned in the treble range, as mentioned, we have decided for now to use the Diffuse Field measurement of our HMS (without modifying it below 9KHz) instead of a Critical Listening room response that Harman suggests. If you look at our review of the JBL Everest Elite 700 (which is based on the Harman Target) it matches our curve up to 1KHz. Sean has said that he wants to include our current curve in a new study that they are developing to see which one their subjects prefer.

    http://www.rtings.com/headphones/graph#447/323

    So the discrepancies that you suggest could be either due to different dummy head responses, our different curve for the treble range, or (as I showed at the end of our video) the fact that different headphones may react differently to different heads (the difference between 600 and 800 S on me at 2KHz was 2dB but on the HMS it was 5dB. And there was a reversal at 10KHz)

    I'm not quite sure yet what role crosstalk plays in frequency response of headphones. My hunch at the moment is that it will not be a big factor, and will only play a role in Soundstage/Imaging. The reason for this guess is that humans are very good at differentiating between tonal and spatial information. As an example, if I measure your HRTF and add spatial information to your headphones based on your HRTF, you will only perceive it as change in spatial information and not the tonality. But if I use someone else's HRTF on you, then you will hear those as a shift in tonality, because your brain doesn't recognize them as valid spatial information. It's definitely something that we'll look into more.
     
    Last edited: Jun 21, 2017
  6. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,589
    Trophy Points:
    113
    Location:
    near Munich, Germany
    I wasn't talking about the bass range, I was talking about the big differences in the midrange. You really shouldn't get such a big 3-4kHz bump with the microphone at the ear canal opening. Did you compensate the readings across the whole range, but with the microphone in the HMS at eardrum level and the other mic at the ear canal opening?
    I generally find mine and @Hands in-ear microphone measurements to be the most accurate up to 5-7kHz. Above that a flat plate coupler gives better results IMO. The midrange up to 5kHz I feel is where your measurements need the most work. Here are a few measurements just to give you an idea. I know you don't have measurements for all these phones on your site, but this is what I expect the results to look like:
    There are more than this, but these should give you an idea. Hands measurements are uncompensated. I now use some compensation to lower 5kHz 3db and to compensate for the upper octave treble rolloff you will see in my older measurements.

    Looks like the dummy heads are quite different. Probably a better idea to make an all-new target curve instead of relying one made in very different conditions. And even then the results might not be perfectly accurate. In this case I don't particularly like the shape of both of these curves, even when compensating for the outer ear gain. The treble doesn't really look all that good, but I prefer the shape of the 45CA. I often see a similar 10kHz peak (and 8kHz dip) in my own measurements.

    Yes, which is why I don't get why there's a 5db bass hump in the target curve.

    I would guess especially the untrained listeners would prefer to have a bass-hump. I have no idea why Harman likes to put a bass-hump in their target response, but they have the same bass-hump in their speaker targets. I hope we can agree that speakers that measure flat in the bass region also sound like headphones that measure flat and that headphones don't need more bass to sound like speakers. Actually I think if anything people prefer slightly less bass with headphones.
    Let's just say I strongly disagree that speakers or headphones should have a bass hump in their response. I don't see how this makes any sense. I know I definitely prefer flat bass. I even dislike it when the subbass is just a tiny bit higher than the midbass. I don't get how a bass emphasis can lead to a neutral or accurate FR. Whether someone prefers to have a bass emphasis is a completely different story.

    I think it's a matter of a compensation curve that could need some work and a dummy head that is probably not 100% accurate to what we hear. But I wouldn't worry too much; your results aren't that bad.

    I didn't mean the crosstalk of the headphones, I was talking more about that headphones lack crosstalk compared to speakers, which I do think affects the tonality. The crosstalk of the headphones themselves is so low that I don't think it matters, not even for soundstage perception. If it did headphones would sound completely different in an anechoic room than they do in a regular listening room, or at least change a lot depending on the room. The lack of crosstalk makes the presentation very different.

    That sounds interesting and I would say matches my experience where I made a few recordings with microphones in my ears and listened to them. The soundstage was far superior to dummy head recordings and the tonality also didn't seem to change as much.
     
  7. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,589
    Trophy Points:
    113
    Location:
    near Munich, Germany
    The more time I think about it the more I am annoyed with the Harman target bass hump. I don't particularly like the shape of the Harman curve in the midrange either, at least based on the graphs Tyll showed us (note how it makes the SR-009 look close to flat above 200Hz). Seriously, this kind of target response makes manufacturers like Focal insecure and they end up trying to voice their headphones to it, just because they think consumers will like them more. I bet Focal could've come up with a better voicing themselves. Just look at how closely the Utopia resembles the Sonoma Acoustics One, which is EQed to a target similar to the Harman curve. I'm not saying they sound alike, I'm mainly just saying they were both tuned to a target very similar to the Harman target.
    But the bass hump. I still can't get the rationale behind it. Shouldn't it be about accurately reproducing sounds? How can the bass hump be accurate? How do you even make an open headphone that is flat, but with a bass-hump without compromising the rest of its range? Just thinking about how a target with a bass emphasis like that could get so far makes me mad. And that probably also negatively influenced the midrange tuning to get to the weird U-shaped 100-2kHz region we have now.

    Actually I just made a quick EQ profile to emulate the Harman target (well, Utopia FR) on my modded HD800 and so far I prefer the flat tuning. The whole 20-3kHz range of the Harman target seems questionable to me. Comparing to my speakers the Harman target seems forced and strained to me. I'll try the same thing tomorrow. Maybe some brain burn in can change my opinion.

    Here are the EQ settings I used in JRiver. Feel free to invert these and try as Utopia EQ:
    570Hz -2.6db Q0.5
    1350Hz 2db Q2
    2000Hz -3.1db Q2

    Because this essentially raises everything above 100Hz, but doesn't really affect the region above 4kHz, you might also add a 2.5db 4kHz high shelf if you want to try this with your Utopia. I did try it with a -2.5db 4kHz high shelf.
     
  8. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,623
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I keep saying these things:

    Doesn't the Harmon Curve only apply to the head/torso apparatus that they used? What model was it?

    Wouldn't different simulators have slightly different responses, so the Harmon curve could not be applied just anywhere?

    Finally I saw a photo of the speaker reference. I recall big JBL speakers shoved against a wall. Speakers that I recall we're already kind of bassy.

    Are there listening position measurements using standard microphone techniques of this "reference" speaker setup?
     
    Last edited: Jun 21, 2017
  9. briskly

    briskly Friend

    Pyrate
    Joined:
    Oct 3, 2015
    Likes Received:
    59
    Trophy Points:
    18
    Location:
    New York
    • The bass boost present with the so-called speaker target is from the combination of room absorption/reflection and speaker directivity, or rather the lack thereof. Speakers lose control of directivity with lower frequencies, and the delayed reverberation is disproportionately attenuated in the upper frequencies. These are all incorporated into the magnitude response if you take in the results over a long enough window. Floyd Toole offers some suggestions about how this might relate to the listener. I have misgivings about applying these ideas to headphone listening.
    • Arguments of "spatial" information and "tone" have been considered before. The brain should have some ability to adjust level based on ITD/ILD of the sound stimulus. This has been a central part of arguments in favor of diffuse field equalization.
    • Interaural crosstalk even for most open backed headphones is low enough to make it a distant consideration when considering the magnitude of the response at the opposite ear. Off the top of my head I can only recall the Stax Lambda series having anything that could be even considered significant.
     
  10. Bill-P

    Bill-P Level 42 Mad Wizard

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    4,801
    Trophy Points:
    113
    Location:
    San Jose, CA
    Huh...

    [​IMG]

    ^ stock HD580 (very similar to HD600 since it's using HD600 grills) versus modded HD580 (same headphone, same setup, etc...). Super imposed and stretched and all that.

    Recessed at 2KHz? Checked. ;) I think it's the "shoutiness" at 3-4KHz that takes away from the 2KHz region directly below it. Also lack of 10KHz is pretty apparent to me. It smooths some songs out but sometimes it makes the HD580/600 sound like they're lacking speed. Basically, everything is related and relative, until you get close to a perfectly flat response.

    Honestly, I don't find HD580/600 "bright" per se but the "shout" is definitely audible to me, as is the lack of "air". And yeah, my mods target specifically those regions.

    Didn't wanna show off the mod or measurement, though. I've been keeping it to myself because I'm just a selfish bastard like that (plus it's fairly complicated). But just to say, you are actually on to something here, man.

    And there is indeed a way to remedy some of it. I kept bass response the same because I'm just a basshead like that. :) (Andromeda sounds to me like it has 1-2dB more bass than this, if you'd like to ask)

    I'm probably one of those folks who'd say the Harman bass response is "just about right".
     
  11. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,623
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    We really don't know what the Harmon curve bass is unless we have the Harmon head torso ear simulator.

    Or is the Harmon curve magical that works for Jude's coupler, Tyll's coupler, my coupler, your coupler, serious' coupler?
     
  12. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,589
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Exactly. In fact my speakers aim to make the bass more directional.
    I think it was down 30db in the upper midrange with the HD800 in a typical room. It's probably much lower than that in an anechoic room. 30db is already a lot.

    Yup and yup. I hear some edge or brightness in the lower treble, but the midtreble balances it out a bit. But I do think these plots don't show enough 1kHz (although it's hard to be 100% sure where it is when there isn't even a scale).

    I don't think the bass region should change at all. Probably very minimal change up to 500Hz or 1kHz from different dummy heads. But yes, can't just use a target made for one dummy head with another.
    They were already kind of bassy, but EQed to have even more bass. I think the speaker target is the one you can see in this article:
    https://www.innerfidelity.com/content/acoustic-basis-harman-listener-target-curve

    One thing that's interesting is that the FR is left flat from 200Hz to 2kHz. I feel this explains to some extent why I don't like the Harman target midrange. Note how the midrange looks far more normal with the 1db/octave slope than with the Harman target.

    Essentially I feel the Harman target is too much like an SR009 with bass shelf. The fundamentals become too recessed. Why not just use something close to neutral like the HD6X0 and go from there?
     
    Last edited: Jun 22, 2017
  13. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,248
    Trophy Points:
    113
    THIS +90000

    We at Sonarworks tried to incorporate the Harman curve and found out that there can be no universal visual representation of the Harman curve. So we had to get creative and... Wait until next week. I'll either announce something or regret this post.
     
  14. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    Thanks for all the comments guys :) I got more than I asked for! Give me an hour or two to write my replies to your comments.
     
  15. briskly

    briskly Friend

    Pyrate
    Joined:
    Oct 3, 2015
    Likes Received:
    59
    Trophy Points:
    18
    Location:
    New York
    The measured curve of Harman's room doesn't generalize if you use a different test fixture. The 45CA isn't shaped in a way that looks remotely like a person, so the speaker/HATS interaction is unlikely to be representative of a person.

    However, the listeners were free to adjust the treble response with a shelf filter, so the more noticeable deviations from preference should be removed by the subjects. The headphones only significantly interact with the pinnae and canals, and can be easily compared to any other fixture that has similarly shaped pinnae. But, the only person I can recall that uses similarly shaped ears is Jude.

    Heads, human or simulated, don't have any sort of meaningful gain below 200Hz. That problem is in defining the speakers and the room acoustics. The initial speaker data used a set of Revel F208s, averaged and flattened over a 3x3 microphone array. The JBLs would have been used some other time. If it helps, Stereopile measured a pair of those Revels.
     
    Last edited: Jun 22, 2017
  16. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    The blocked ear canal microphones (for short mm: miniature microphone) were calibrated against the ear simulator with the pinna removed. This was the only way they gave proper results when mounted on humans. But calibration or compensation really doesn't make a difference in the point I made in the video. Since only the relative difference is the issue here, and not the absolute difference. The point was that two different headphones, react differently to different heads. Here are a couple of freshly made measurements with the miniature mics just for you guys (top 3 measurements are the HD 800 S, and the bottom 3 measurements the HD 600):

    [​IMG]

    Take a look at the relative difference between the 3 heads. These were all measured with the miniature microphone, even the HMS measurements. So the only variable here is the head shape/size. This is not HRTF, this shows that different headphones react differently on different heads. Also note that I'm getting less Bass with the HD 800 S than the other 2 heads (and more bass with the HD 600!).

    Here is another view of the same measurements, arranged by person (Green: HD 600 | Red: HD 800 S):

    [​IMG]

    In the graph above also note the difference in the Bass Range!

    Yes, one of the reasons the Harman curve didn't work for us, was that it was based on measurements on a different head. That's why we made our curve based on the Diffuse Field measurement of our HMS. So I think in the end the Headphones Target Response will be more of a "standard" than a fixed curve, since each head and torso simulator (HATS) should have its own target curve.

    It seems that there's no absolute target curve nor an absolute headphone frequency response, unless they are measured and calibrated on and for a specific individual. Otherwise, the best we can do is to do an average of an average of an average! That's why self-calibrating headphones are the way to go!


    The Harman curve is based on the assumption that the perfect headphone sounds like the perfect loudspeaker in the perfect room, whatever perfect means. And this was largely confirmed in a few studies where listener's preference was measured. Harman took Floyd Toole's research for loudspeaker target response as the starting point. Toole's research wasn't really earth shattering, in the sense that it didn't contradict the already accepted consensus that the ideal loudspeaker measures flat in an anechoic room.

    Yes, it seems that headphones need 1-2dB less bass than speakers to sound similar (based on Harman's research, which was confirmed by my own tests). This is the Harman In-Room response in blue done with a measurement microphone:

    [​IMG]

    The red line is the in-room response of the Sony A1E TV, done with no EQ. The bump you see from 200Hz to 400Hz is the due to the room modes. That's the only thing I would EQ out with this TV. The bass sounds just fine.

    *EDIT*: I agree that using a TV frequency response to support my claim was not the best idea. But I have listened to a lot of TVs recently, since I'm developing the sound section of our TV reviews as well, and this is a very good sounding unit, for a TV of course. And its In-room measurement is very close to Harman's In-room measurement. Another TV with great sound which measured even closer to Harman's was the Sony X930c (from a couple of years ago). My point was that more and more manufactures are building sound units based on Floyd Toole's research and to my ears, they sound better than other designs.*/EDIT*

    In terms of my own personal opinion, which could be wrong, I find the Harman target for loudspeakers to be perfectly fine, and I don't see why one would prefer the 1974 B&K curve for example. But putting subjective impressions aside, Floyd Toole's research has been validated many times in double blind situations. My only guess about the reason for the bump above 300Hz in the 1974 B&K curve is that it could be a post-emphasis filter. Since older recording mediums did not have enough bass/mid, this curve would add 'too much' bass/mid to the playback in order to balanced it out (but even this idea is a bit of a stretch). Below, you can see Harman's In Room response and rtings headphone target response. So yes, just a little bit less bass on the headphone target.

    [​IMG]

    In my opinion the effect of crosstalk on headphone's tonality is also minimal. Unless it's done poorly, then you hear a lot of comb filtering. But if done right, it just pulls the stereo image a bit together and makes the stereo image more narrow. In my listening tests it doesn't really affect the tonality enough that would explain the difference you see in the graphs.

    The bump is there because if you EQ your speakers to sound flat in a room, it'll sound too thin. Mixing/Mastering studios, if done right, won't measure flat. They show a bit of build up below 250Hz, just like the Harman In-Room response. So if something is mixed in a room like that, your headphones should also have a similar bump in order to reproduce the tune as the engineer intended. I don't have access to the headphones you mention here, but the HD 800 S which measures flat uncompensated below 500Hz, to my ears and to the ears of my colleague here lacks about 4dB of bass at 20Hz. The lack of bass starts to show around 90Hz (check out our measurement on our website). If we can't agree on this, then we definitely don't hear the same thing, or don't listen to the same music :)

    Yes Marvey, it was a GRAS 45CA. I believe I have answered the rest of your questions earlier in this post.

    Yes! The crosstalk on open-back headphones may have a small effect on the sense of space (Soundstage based on rtings terminology), but none on tonality or other things. It's neither loud or correlated enough to make a dent.
     
    Last edited: Jun 23, 2017
  17. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    Let me know if I've left any questions unanswered
     
  18. briskly

    briskly Friend

    Pyrate
    Joined:
    Oct 3, 2015
    Likes Received:
    59
    Trophy Points:
    18
    Location:
    New York
    Not a sound quality concern, but a functionality one. I can't find impedance measurements of the headphones tested. Nor can I find the sensitivity. Where is that information?
     
  19. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    Thanks for the suggestion. It's on our to-do list to have a Impedance and Drivability section with all the relevant information in there. But it's not our highest priority task at the moment.
     
  20. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,589
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Yes, it is. I meant that the crosstalk influences the tonality with speakers. Since you don't have that with headphones the tonality will still be different, even if it measures the same. So I don't think I agree that headphones should simply measure with the same FR as good speakers to sound the same. The soundfield is totally different. But the approach as to how to get to a target response doesn't really matter ... the results do. Any if I don't think a target sounds good, then it doesn't really matter why. With a different speaker target curve, like the B&K target, I would probably like the Harman target.

    Hmm. I know I don't want the measurements to show a bass emphasis, otherwise it'll sound too bassy. And I know I'm not the only one here who thinks so.

    If anything the Harman target adds too much bass. I don't see how the B&K curve has more bass. More energy around 300-800Hz, yes.
    Yes, if. And if it's mixed in a room with a curve similar to B&K or a general 1db/octave curve it'll sound bassy on stuff with a Harman target FR.

    Oh I don't disagree that the HD800S lacks 4 or 5db at 20Hz, maybe even more so. But I don't agree with the shape of your curve. Your graph makes it look like the whole bass is missing, while I hear a small 100Hz bump and from there on it starts to drop off. Like this: https://www.innerfidelity.com/images/SennheiserHD800S.pdf
    The HD600 and HD650 have even more of this mid-bass hump. This viewpoint is generally accepted in the headphone world, so I don't really see why Harman suddenly tries to redefine that flat bass actually means a 4db bass-hump below 200Hz. Planars generally measure as a straight line below 500Hz and they sound like it too, not like a sudden drop that starts around 200Hz.

    BTW: Using the TV with acoustic surface technology as a reference as to how speaker bass is supposed to sound like ... o_O
    The TV isn't even a good tonality reference since it's mono and speakers are stereo which influences the FR a great deal. And frankly when I heard it in some electronics store, it sounded like shit.

    Do you think your target results in a "perfect" FR or do you think it can be improved somewhere. I don't see the issue with making your own target when it sounds better to you. This is why I told you where specifically I thought your target could be improved.

    Your measurements:
    Were they both in the same position or was one microphone at the eardrum level and your mics were at the ear canal opening? I just don't see why the 5-6kHz region is so high when me and Hands get very different results with our own microphones. Where are the mics positioned? Inside the ear canal?
    Me and Hands insert the mics to be flush with the ear canal opening, as if the ear canal wasn't even there. These results IMO need no calibration to look like calibrated dummy head measurements.

    Looking at your measurements I don't get how the bass on the HD800S can measure flat. It never does, so something is weird.
    What's interesting is the big differences in the treble. I think I have measurements from 4 people and they generally looked very close, with some very small differences at 2-4kHz (differences that can be explained by slight positional changes). The treble generally followed a similar shape, but the 6kHz region was one area where the results showed big differences. Some people had 10db more at 6kHz (regardless of HD800 or HD600) than me!
    Now the question is if it's a matter of repeatability, or if people really do hear things this differently.
     
    Last edited: Jun 22, 2017

Share This Page