1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

2017-18 Harman IEM Target

Discussion in 'Sound Science' started by HuoYuanJia, Jun 4, 2018.
  1. HuoYuanJia
    Keith Howard from IF just announced the new IEM target response: https://www.innerfidelity.com/content/harman-tweaks-its-headphone-target-response

    I am generally not a big fan of the Harman target. I think it started off great and it was based on research and logic. However, their "method of adjustment (MOA)" makes it relatively useless for us. Effectively, it's tuned by mass consumers including damaged and untrained ears.
    The whole approach to "high-fidelity" seems to have gone amiss.

    I vectorized the target and put it in a more common scale.


    I think the bass quantity is very high. It is also pulled down far into the sub-bass. Either this is what people thought would give the best physical impact or it is an MOA error. Users could adjust bass and treble, but not freely. The anchors were set by Harman. So if somebody was lacking warmth, he might have exaggerated with the bass switch and turned it higher yet not achieving the result he actually wanted.

    Same goes for treble. It looks very bright. It is hard to believe that all of the users had healthy hearing. Also, I do believe that the volume was set by Harman and not user-adjustable. It looks a lot like if some users wanted to boost the volume by boosting lows and treble.

    I have overlaid the target on some decent measurements (industry standard IEC 711 coupler).


    Here's the beloved Andromeda. In comparison it is super warm. Not just a bit warm, but like really, really warm. By Harman standards the Andromeda sounds full and muddy. The upper midrange and lower treble is lacking up to 10 dB. Overall, it doesn't quite match user reviews. (Well, it kinda matches my perception, though not that severe.)


    Here's the InEar ProPhile 8, seriously one of the most balanced and neutral sounding IEMs I ever heard. Subjectively, this is the closest I've experienced to the tuning of studio monitors. Well the PP8 does have a +3 dB bass switch which puts it closer to the target, but I actually think the boosted bass is distracting and fatiguing. I prefer the neutral setting. I find it interesting that some people even claim the midrange were too forward and even "shouty". I really can't agree there, but if it were in accordance with Harman, the vocal presence could be pulled forward quite a bit more.


    Here's another TOTL in-ear which recently starts to gain in popularity. The qdc Gemini is considered fairly neutral. In my review I described them as mildly warm for a bit more enjoyment when compared to studio sound. I was referring to the lower treble and a slightly early cut at 3k. Yet some users describe them as bright. The Gemini also have a bass switch which would bring the lower pressure closer to the target. Again, too much for my taste unless I'm in a loud environment.

    Of course it's difficult to compare the target if the used equipment is not exactly the same. I am not sure which coupler was used to create the target but it shouldn't be too far off. I've already thrown my salt into it; now I'm interested in what you have to say.
  2. castleofargh Contributor
    as you start with a copy of a copy of the target, then directly compare it to IEMs measured by different gear, you can indeed throw a few bags of salt in your interpretations. ^_^

    Harman was very clear from the start that their aim was the preferred target response. it started with the premise the response at the ear of a listener using flat speakers in a room would would be the preferred sound on headphones/IEMs, whatever. the idea coming from their own research on speakers finding that the preferred response was pretty much flat speakers in a room. they started with a fairly objective approach to get that signature as a starting point, and when studies showed a variations, instead of trying to confirm their assumption, they went with what people preferred.

    I used to complain about the first big paper and the 2 sliders(+/-bass and +/- trebles) as EQ for what seemed like the main test. but they did a lot of experiments since and kept trying to zero in on the average preferred signature. so I'm not sure this is a valid critic anymore.

    the IEM target has more bass, which has always been my personal experience when trying to get close enough signature between IEM and headphones, maybe a full size headphone somehow vibrates a little and gives a small tactile aspect to the low end, or something like that, TBH I don't have a clue. but we're a bunch of IEm users needing some extra low end compared to full size headphones. I'm very fine with electrically flat on open circum-aural gears. I could never settle for electrically flat low end on IEMs.
    how much would obviously be variable, if only because listening levels and our equal loudness contour.

    the 5 to 10khz people settles on seems more present than the traditional ideas of almost constant roll off starting around 3khz. maybe this is objectively justified by interaction, or lack of with the concha when using IEMs? in any case they settled on this by using listeners' preferences, so I see no reason to doubt that the listeners did in fact prefer that. I personally get rapidly annoyed by what I feel to be extra 4 to 5khz in IEMs so I always EQ that area in a way that doesn't seem to follow any standard of any kind. but then again if I learned something from 3D simulations solutions, it's that my HRTF is not average.
    anyway, beside that 5 to 10khz area and the low end boost(that has been suggested for a long time even only as a way to substitute for lack of tactile bass), it's not a huge departure from various attempts to get an average neutral target. personally I'd rather have this than classic roll off and some spike around 10khz to pretend like the trebles are good, like I've seen and heard on so many BA IEMs.

    I'm a little sad that this time, the all thing isn't on open access at first, like other stuff have been in the previous years. I'm a poor guy and the AES price for a paper makes me question my level of curiosity ^_^. but I sure would like to know the actual reference for the curve before I try to just apply the variation to my own raw measurements and pray that it's kind of close enough. well I'm going to try off course, but not knowing much of anything will make it hard to draw any sort of conclusion for myself. on the headphone curve, my own preference was almost right in the middle between DF and the Harman target. I suspect it will be about the same for this one, based on my usual EQ for IEMs.
    hakuzen likes this.
  3. bigshot
    Balanced speakers in a room is a great thing to shoot for, but when you're dealing with little tiny speakers you shove in your ear canal, you have to assume that whatever you come up with is going to be an approximation. The room adds a lot to the sound.
  4. SilverEars
    I really don't understand why some people keep using the word neutral and PP8? I don't hear neutral and Andro is not super warm.

    This is just expectation bias from graph, and trying to rationize as warm.

    Personally, I didn't like PP8 all that much. It's too lower treble articulate, in which people associate with more detailing, which is not more than Andro. There's a difference. I also feel it's lacking mids.

    I prefer the Andro response weather or not the graph above is accurate. I could care less if it don't LOOK "neutral" on the graph. It's all about my hearing preference.

    Anyway, please learn that looking at graph and target curves and "neutral" bias your brain think there is neutral there. And people have different idea of neutral.
    Last edited: Jun 7, 2018
  5. HuoYuanJia
    @SilverEars I don't think you really read my post.

    If you don't think of the Andro measurement as accurate, feel free to compare it with Ken Ball's measurement, also posted here on Head-Fi. Looks very accurate to me. I compared the Andro curve to the Harman IE target, based on that the Andromeda are " super warm", my personal "bias"/opinion is only put in brackets.
    You think the PP8 have too much upper mids/ lower treble, that is fine. I think it's accurate but I also wouldn't want any more energy. However, Harman's research suggests that another 5 dB were for the preference of the majority. So we both agree.

    So while you tried to express that you have a different opinion than me, you actually supported my concerns. If anything, it seems you agree even less with the target. (You seem to think the Andromeda curve should be a reference target for IEMs?)

    BTW, it's AES paper 9840 (and follow up 9878). Sadly it doesn't mention anything about the measurement setup (or I forgot) but it is hard to imagine they would use different norms than the industry standard so the curves should be at least comparable. If not, what would be the point in releasing them? It seems Inner Fidelity will match their measurements and apply the Harman IE correction in a web tool so of course we should be able to discuss and compare the target.

    I applied the curve via EQ to an IEM. It sounds very airy. It reminds me of the Audeze iSine 20 with Cypher cable. I find it very, very fatiguing as voices shout a lot. But surprisingly it handles sibilance far better than I would have thought. I can see that somebody would like the target. The sub-bass presence is "wow" without making the IEM sound warm or even too bassy. The bass quantity doesn't really come through that much because the treble is pushing forward even harder.
    What I can say already is that the mind adjusts quickly to the bass. Deactivating the EQ makes me believe I have lost a lot of information and it will take a while until I can hear the bass details again. Truth be told, all in all, the Harman target sounds far better than I ever expected. Hell, I think I could even get used to it.
    However, I think it is time that somebody comes up with an alternative target, a more "correct" one for prolonged listening sessions. The Harman curve seems more suitable for a quick fix while waiting for the bus.

    If anybody wants to try it, I found equalizing the PP8 to it was the easiest. Here is a simple preset that works with the RME ADI-2 DAC/Pro and has a tolerance of less than 2 dB:
    1. G +12 F 26 Q 0.5
    2. G -3 F 29 Q 1.0
    3. G -1 F 230 Q 1.0
    4. G -2 F 1.25k Q 3.0
    5. G +7 F 3.5k Q 1.0
    Last edited: Jun 7, 2018
  6. SilverEars
    Stop insinuating what I have not said. You are free to your "opinion" as I.
    Last edited: Jun 7, 2018

Share This Page