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

Is there such a thing as a flat IEM?

Discussion in 'Sound Science' started by Killcomic, Dec 27, 2019.
  1. Killcomic
    Right. So, I bought the ER3XR after people kept creaming their pants when talking about Etyimotics and their IEMs.
    They keep getting referred as 'flat' and 'neutral'.
    I'm not massively knowledgeable about how the inner ear works, but my understanding is that a flat frequency response does not sound flat to you inner ear, so actual sound reproduction needs to be done with a totally not flat response in order to sound flat.
    Problem is, how do we know it's flat?
    My ER3XR sounds very mid forward, but not artificial. However, I don't know if it actually sounds natural or it's just my ear perceiving this.
    Can anyone shed some light into what an actual neutral/flat/pancake IEM sounds like?
    Do I perceive the ER3 as mid forward just because all other headphones/IEM are have recessed mids?
     
  2. sonitus mirus
    The frequency response on all the tests I could find seem to indicate these IEMs have an elevated hump up to 8-10 dB in the mid range.

    https://crinacle.com/graphs/iems/etymotic-er3xr/

    Seems like your hearing is just fine. Flat and neutral compared to what, other IEMs? I'm a Denon AH-Dxxxx fan, so practically every other headphone seems mid forward to me. :ksc75smile:
     
  3. bigshot
    The problem there looks like it isn't so much the midrange bump. That isn't far from the Harman curve. The problem is that the bass is flatlined. If there was more low end, the upper end wouldn't seem so pronounced. All bands are relative. Too much of one may actually mean too little of others.

    Headphones shouldn't be a flat line. Speakers in rooms shoot for that, but the room adds to the sound as much as the speakers do. "Flat" should mean "balanced", not necessarily a line on a chart. That is why graphs are often compensated.

    Headphones should follow the Harman Curve as a baseline. Then feel free to EQ a little bit here or there to bring them into what sounds right to you. IEMs add a whole other layer of variability because you're shoving them in your ear canals. There is no way to predict exactly how they'll sound without trying them. If they don't sound right, EQ.
     
    Last edited: Dec 27, 2019
    bjk8kds likes this.
  4. castleofargh Contributor
    There is no flat FR reference for IEM(or headphones for that matter) because variations between people are expected and found. It's that simple.
    If those variations were tiny and mostly unnoticeable, we would settle on a definition of neutral and move on. But it's not the case in practice. Some variations can be very obvious between 2 listeners using the same IEM or same headphone.



    Let's say your outer ear has a shape that boosts the 6kHz area a lot(more than the super average human ear does). When you insert the IEM, you bypass that boost you have learned to consider normal and neutral. So if the IEM only boosts the 6kHz area the way a perfectly average ear(or a dummy head) would, you will subjectively feel like the IEM has slightly recessed 6kHz.

    Another of many similar examples, ear canal length: let's say the average dude has an ear canal that's around 2.5cm. If you happened to have a rather big head with a 3cm long canal, assuming everything else equal(which probably isn't the case), that alone could shift the main ear canal resonance down by more than 500Hz. While a tiny ear inserting that IEM as deep as anybody else, would most likely need the "3kHz bump" centered above 3Khz(maybe 3.5 or something) to feel the neutral midrange tuning that person is used to when listening to real sound sources around him/her throughout the day.
    When you insert your IEM, you block quite an area of the ear canal. The remaining space might resonate at maybe 6 or 7Khz, while the natural open ear canal was resonating around 2.5 or 3kHz. A so called "neutral" IEM must have taken that into account when tuning the IEM. But if it did it for some ultra average ear, the values will be shifted compared to your own perceived flat. How much and how noticeable noticeable will depends on who's using the IEM.

    And of course, depending on how you insert the IEM, your obstructed ear canal will not resonate at the frequency the "flat" IEM predicted it would. If you insert the IEM like anybody else, with your canal being longer in my example, the remaining space will also be bigger, and the resonance will exist at a lower frequency compared to average human ear predictions.



    If you keep looking, you will keep finding reasons why different people hear differently with IEMs. So a universal concept of neutral can be either arbitrary for the sake of having a reference on a graph, which is something we do need to make sense of FR graphs. Even if it doesn't sound flat for most people(kind of what diffuse field compensation has been for a few decades). Or be some averaging of neutral for many listeners, meaning it will effectively sound neutral only for a portion of the population.
    IMO we'll only be able to settle the matter of flat the day we get better tools for customized responses. Ideally they'd be based on some HRTF capture of each listener, but I'm open to more creative or subjective alternatives. Until then, a neutral IEM is a subjective concept and a nice EQ is your best friend to help find your own neutral.
     
  5. gregorio
    1. True.
    1a. Not true! It's not true because you're missing a vital part of the equation, which is: What is the freq content of what you're trying to reproduce (the audio recording)? For example, if we took an audio recording of white noise (which has a flat freq content) it will not sound flat, it will sound rather trebley because human hearing is not sensitive to low freqs. However, you're ignoring the fact that our audio recording has been made by humans (musicians and engineers) who of course also have human hearing and therefore if they wanted it to sound flat they would have increased to bass to compensate for their (and your) insensitivity to bass freqs. What would therefore actually be on this recording would sound flat but actually not be flat, it would be more similar to pink noise than white noise. When it comes to reproducing this recording, you would indeed want a system with a flat response, so that it accurately reproduces the frequencies on the recording (which are NOT flat)! In other words, the compensation for the insensitive/sensitive areas of human hearing has ALREADY been applied to the recording before you try to reproduce it (automatically by the musicians/engineers, because they too have human hearing) and therefore you should obviously NOT try to apply compensation again. The situation is more complex with IEMs because you are changing/bypassing the response of certain parts of your hearing, for example by eliminating your pinnae and changing the effective length of your ear canal, as castleofargh explained, and therefore some compensation might be required, depending on the individual.

    2. We can measure their output. However, their output is never flat because headphone manufacturers always apply a non-flat response to compensate for the sound source being right up against (or in the case of IEMs, actually inside) the ear. Theoretically we could measure the actual sound at the ear drum with a probe but I'm not sure if that's possible/practical because the probe (or it's wires) might affect the IEM's seal and therefore the freq response at the ear drum. In practice then, we don't know if it's flat, all we can do is make a guess based on a comparison with a reference.
    2a. The last part of my last sentence is particularly problematic in the audiophile world because exceedingly few audiophiles have an accurate reference. In effect, their reference isn't a reference it's a preference, based on; some expensive audiophile system they once heard, what they think their music files should ideally sound like or a combination of the two. Without exception, every audiophile I've ever had in my studio has been surprised/shocked, including audiophiles with decades in the audiophile world and many tens of thousands of dollars spent. In short, there's no easy way for you "to know". If I were you, I'd consider them to be perfect for the time being and just enjoy them without trying to make too many judgements, then in a few months time, when your brain/perception is fully acclimatised to them, decide if you really want to apply a little EQ.

    G
     
  6. castleofargh Contributor
    I left albums out of the picture because then we fall right back into Floyd Toole's circle of confusion and my post was already stupidly long and boring for what I had to say. But of course it is part of the real situation in music playback and deserves to be mentioned when discussing flat.
     
  7. Killcomic
    Oh, man. What a rabbit hole!
    I guess what I’d like is to find something that’s true to the source recording, but it appears that is far harder than I initially suspected.
    Perhaps, if I’m not a sound engineer, I’m better off simply finding something I like and just enjoying it without thinking too much about it.

    Thanks guys!
    As always, I’m a bit humbled at your level of knowledge on this matter.
     
    Last edited: Dec 29, 2019
  8. bigshot
    The trick with these sorts of things is to pay close attention to the broad strokes, and don’t sweat the details. Get in the neighborhood, then fine tune by ear.
     
  9. castleofargh Contributor
    The main difficulty is really to get yourself a flat speaker reference as @gregorio mentioned. But if you forget about some dream of perfection, you can use a reasonably balanced speaker as your reference and EQ the IEM to closely match that FR. @Joe Bloggs posted a tutorial some years back on how to use your own equal loudness contour as intermediary between the IEM's FR and the speaker's FR(or any other source you wish to take the FR from). It's kind of a PITA and in my case, having little experience of EQ at the time, it took me many trials to get something consistent. But it worked great in the end.
    Similar process is explained by David Griesinger here:

    And while it was some time ago, he readily gave the little app to ease up the process in reply to a request by e-mail. That app when I tried it wasn't perfect, and it was relatively easy to clip the signal depending on where I would start in frequency, but otherwise it removes a few steps of Joe's tuto.
    Griesinger's explanation and decision to use only one speaker right in front for the calibration is because he does this specifically in the mindset of binaural recordings. So all he wants is to calibrate his own perceived vertical axis and to do that he doesn't even need a very flat speaker. But obviously you can do the same with a pair of speakers and get your head related correction from sound sources at 30° on each side, instead of like he does, getting the head related correction for a sound source right in front of you. Both have meaning but the typical stereo model is still stereo speakers for most albums. So that's what we would like to have as the flat reference IMO. Again the main obstacle beyond personal efforts in a bunch of trials to get it right, is to get a decent response at your listening position with a pair of speakers. Not everybody has that.

    To be clear, the FR is only part of what needs tweaking if your aim is some unicorn idea of the sound like the artist intended. Starting with how obviously the IEMs aren't sending the left channel to the right ear like a left channel speaker would. Then room impact, tactile bass, sound changing with head movements, knowing the actual listening conditions for a given album at a given studio, ...
    In comparison, aiming for a given frequency response is the easy part for us :sweat_smile:.
     
  10. Killcomic
    I just need to stop fearing the EQ.
    I was always in the mind that if I had to EQ a headphone, then I bought the wrong headphone.
    The more I learn how people hear things differently, the more I realise that I was just robbing myself of a more pleasant listening experience.
     
    BobG55 and old tech like this.
  11. bigshot
    There is nothing wrong with buying headphones that come close to your particular target. I have ones that are 95% of the way, and that is good enough for me to use without EQ.
     
    BobG55 and Killcomic like this.
  12. castleofargh Contributor
    Most FR tuning is effectively EQ even if it doesn't have that name and doesn't show little sliders. If an IEM has some circuitry, multidriver design that massively band limits some driver, or just like many Ety IEMs, a pair of resistors to change the signature a tiny bit, all that is 100% EQ!

    The real problem of EQ is that it's not as intuitive as most people think. And also that many DAPs still have really limited and sometimes plain crappy EQ. But the idea of EQ itself is neither good or bad, it's a tool for when we need/want it.
     

Share This Page