Only because it had way more bass… Obviously it would be preferred over flat DF.
Nobody is advocating for flat DF. This is an extremely disingenuous argument that we more typically see coming from obfuscators. I really hope this isn’t actually what you’re arguing.
The appropriate point of comparison is DF with preference filters.
You could argue the same thing in reverse but in-room flat vs DF with preference filters is similarly not a meaningful point of comparison.
Huh, interesting, this made something click for me that never clicked before.
But before I get to that: in the above I wasn’t talking about any brain judging anything, only about the HpTF variation graph and what information it does or does not contain, based on how it was captured and what corrections were applied to it afterwards. But this might be the wrong thread to be digging into that further. For me it reopens a “line of questioning” I had with you guys last year, that I dropped after a live section of it stalled on the podcast because I couldn’t type my replies fast enough and you guys moved on. I’ll maybe pick that up where I left it off in the forum.
Now wrt. the brain expecting DF… I do remember that diagram of the stages of auditory processing where we first have to match the spectral profile to some “memory bank” profiles that correspond to spatial directions, and when headphone sound doesn’t provide enough directivity cues the brain defaults to some directionless spectrum, which I guess is pretty much proven(?) to be the listener’s DFHRTF.
But that gets me thinking… since everything about complex responses of social and biological systems is probabilistic, not clear-cut and binary, maybe there are personal differences in how sensitive we are to the different directional cues and how realistic of a set of cues is needed for our brains to activate a directional map and not default to the DF. Like maybe you DF-ists have sort-of “fanatical realist” 
brains with a super high bar for directional cues and thus have a far stronger tendency to default to the DF map, whereas we SRF-ists lean more toward “please tell me beautiful lies”, more eager to be transported/immersed/convinced by imperfect simulations, and thus more easily activate our frontal-direction spectral map even when the sound doesn’t contain perfect cues (like when only the FR fits but not all of the ITDs and ILDs down to the last detail).
And if that’s the case, it could have something to do with you DF-ists having too much training with too many test materials, too many reproduction setups, and knowing too well what the differences are between real sound and imperfect reproductions of it. Could be interesting to separate people’s reactions to the OAE1 based on their level of experience with analyzing audio by listening to a variety of gear, or based on their history of professional vs. consumer engagement with audio tech. Some of you DF-ists might be living inside a certain “uncanny valley” where the OAE1-like response seems wrong to you, while to most naive listeners but also to extremely experienced engineers like Grell, it comes out very close to natural and thereby enjoyable. 
Not sure about the memory bank bit but yes, the latter portion of this is what we’re talking about. As for proven, listening tests have been done on this, decades ago. DF has also been established as the international standard for headphones.
Something along these lines is Axel Grell’s conjecture. But really we need to think of the front-biased timbre mapping as ‘tricking the brain’, given there are no locational cues when you’re wearing headphones the way you get when naturally hearing the world.
So yes, headphones are the unnatural listening condition, however creating a front-biased sound field within that condition requires tricking the brain to interpret sound as if localization cues existed in that condition in order for the timbre to not sound strange.
Mark Ryan brought this up on the live stream recently but it’s a bit like trying to do one of those magic eye illusions where there’s a hidden 3D image within a 2D picture. That’s sort of what your brain is being asked to do with front-biased sound field conditions in headphones. For some of us, that may be easier to do than for others.
Not a single one of us is advocating for flat DF as a preferred target response.
Since down-tilted DF was never tested vs. down-tilted SRF in any of the papers, I have no idea what claim you think exists re: Harman justifying SRF over DF as a baseline to add preference adjustments to. There is no basis for this.
The need for personalization is the core conclusion of Harman’s work. Even amidst everything we’ve learned since it was published, the need for personalization due to the plurality of preference still remains the most salient, consistently-shown-across-papers conclusion. Every single time I see Sean in person, our discussions come back to how important this is.
Frankly, I cannot fathom an argument against personalization, it goes against literally every single listening test conducted in all of Sean’s work; literally every subject preferred something different, so listeners should be able to adjust the sound as they please. I’m hoping I just misread you here, frankly 
Have you read Thiele’s paper on the subject, by the way? There’s a reason Sean now uses DF as his pre-adjustment baseline; if the theoretical argument is compelling enough for him I don’t know why it wouldn’t be compelling enough for the people who follow his work unless they haven’t earnestly engaged with the theory. Not only did he take measurements in the same IEC room to show the difference between DF and the original In-Room Flat had nothing to do with anything but the amount of speakers used, but it was enough to convince him that moving to DF instead of a 2-channel quasi-HRTF actually makes sense given headphones are devices of Diffuse localization.
I don’t think EQing your headphones to an average curve of a bunch of other headphones that aren’t actually tuned to the Harman target is engaging with any part of Harman’s work. I get that you’re probably rationalizing it by placing them against a Harman target and seeing adherence, but you’re EQing to a response that, sure, resembles a downsloped HRTF response of some sort… but you’re basing it on specific headphone/ear load interactions—none of which were present in Harman’s work—which means, if anything, you’re a good few steps further away from following the Harman work than you think you are.
Ask yourself: is what you’re hearing really what the subjects heard? If the answer is “yes,” I have some bad news for you…
It doesn’t really matter for your own personal use, of course—obviously you can do as you like—but I think if you’re trying to frame your actions as somehow being more in-step with Harman, you seriously need to take a step back & recognize the fact that what you’re actually listening to likely sounds meaningfully different than whatever Harman’s subjects heard. Why? Because headphones are load-sensitive and the ears used to dictate the target response (and measure the headphone measurements you’re using/averaging) fundamentally misrepresent the effects of the ear used in the research, as well as the geometry/impedance of your own ear.
I get that you’re getting results you enjoy, but that alone is the justification for your methodology. To my mind, it’s sufficient for your enjoyment and that should be all that matters, but Harman’s work doesn’t actually have any supporting basis for what you’re doing, and furthermore, when diving deeper into Harman’s work, it shows that what people have basically been EQing to for years and calling “Harman” isn’t actually all that close to what the original subjects heard.
Instead of talking about what’s right or wrong I tested the tone generator on my HD600 as well (did my Mirph1 above) that I had eq:d with the suggested adjustments published here. For the most part they suited my ears apart from the lowest bass 30-40HZ that were a bit low in level as well as 7000Hz that also need some boost for me. The rest is sounds pretty flat to me 
. Again10 minutes work and a free improvement.
Mike
I learnt a lot from the main Youtube video that underpins this thread. I generally prefer IEMs cos they are not expensive, especially those made in China by Chinese brands. Mine is the ARTTI T10, a budget $65 IEM - planar magnetic.
Long story short, I tried using Auto.EQ, which is an online site dedicated to creating EQ correction filters, based on headphone/IEM measurements stored in its database.
Based on @GoldenSound ‘s video, I then made my own EQ filters further to listening to the tone generator highlighted in the video.
https://www.szynalski.com/tone-generator/
I preferred the results of my own effort, created from my own listening, which compensates for
If you are interested the full details are here :
I love seeing more people adopt the sweep and listen for peaks and dips method! It sounds like many of you have had a positive experience doing so, which is awesome.
Still wondering about this. @Resolve? @listener? Any feedback even if it’s “I’m not sure. Give it a try and let us know.” would be helpful.
Sorry, hadn’t seen it. In general they’re useful for creating fun binaural recordings and if you can somehow get an HRTF with that microphone I’m sure it’d be better than nothing, but not sure if the raw results would give you much useful data given how far out into the pinna the mic elements seem to sit.
What I really don’t understand is how an EQ change is supposed to correlate 1:1 with measurable/audible output. If headphone X produces Y dB at frequency Z on head 1, and Y+3 or Y-4 on head 2, why is it that we believe adding +5 to that (put Q to one side for the moment) will result in +5 to both head one and head two? Shouldn’t it be more likely that it would come out differently?
What’s the easiest way for someone with a pretty modest setup and modest experience to give this a go? Is there a relatively accessible site that allows a slider back and forth across the frequencies?
I like this website.
Headphone X will produce a measurable 5db louder sound at Z frequency regardless of which head.
What comes out differently is how Head 1 or Head 2 perceives Z frequency.
But even just taking Head 1, there’s also positional variation to take into account on how Head 1 hears Z frequency, and the higher the acoustic impedance of the headphone, the more affected by position.
Lots of variables for how even just one specific headphone (to take away the variable of unit variation) sounds different for different people, and why personalized EQ is so important if you want to maximize the “Sound Quality”.
Thank you for your well-considered and thorough reply to my posts here listener. We obviously look at some of this a bit differently, including some of the Harman conclusions.
Since you have occasional interactions with Dr. Olive, you might try asking him sometime if he still believes that most headphone listeners prefer a response close to the in-ear response of neutral speakers in a semi-reflective room. And if this is still one of his principal takeaways from the research he did with Harman. I’d be curious to know the answer to that. Because he has said as much in a number of interviews and talks he’s given about his work there. (Three of these talks can be found here.)
I’m also glad to hear that you’re not advocating for DF as a target. Because it often sounds like you believe that it should be when you simply refer to your target or baseline as “DF”. After reading your remarks above though, it’s a little clearer to me what you mean by some of this. A flat DF (or just “DF” in most audio circles) is not what you want to be using to judge the timbre of headphones though. Because it is simply too bright. I hope that most of us can agree on that. (Though I know of a couple people in your inner circle who still seem to think it should be the target, or who are at least open to using headphones that are close to DF.)
If you want to use DF as your starting point, then you obviously need to make some adjustments or corrections to it to bring it more into line with what most people would prefer, or would likely find more pleasing or neutral-sounding. IOW, you need to make it closer to the in-ear response of neutral speakers in a semi-reflective room. In my book, that means the DF is no longer a DF anymore. Because you’ve changed the timbre or tonal balance to match something else, namely the in-ear response of speakers in an SR room. And presumably this now becomes the new baseline, or reference point, or target, or touchstone, or whatever you want to call it, for judging the frequency response of your headphones. And (potentially) also making some EQ adjustments to correct it.
If you add a negative slope or the Harman filters to a diffuse field measurement to create your target, then the diffuse field measurement is no longer “DF” anymore. It’s DF plus (or minus) something else. And more than likely that DF+something is gonna be closer to something like the in-ear response of neutral speakers in an SR room, which is the reference point or target I’d really prefer for headphone measurements.
I also understand why alot of reviewers don’t want to hear something like that. Because doing good in-ear measurements of neutral speakers would be a costly and involved process… I get that. But it’s something I’d still like to see people try to work toward, instead of just more messing about with DF and different filters… Anyway, thanks for letting me say my piece on some of this.
I think the subject of personalization is also interesting. And maybe we can also have a more in-depth chat on that sometime too.
I’m just not convinced it’s that important for the kind of over-ears I use. And haven’t found it very effective in delivering the kind of results that I want. I can see some value in it for IEMs though. I’m afraid for many people it will be another unicorn chase though… That’s the way it felt to me anyway.
In your post above I think you also mentioned that the Harman target that we used to know and love wasn’t really the real Harman target. For those here who may be interested, there are some recently revised versions. Though they are still quite course, and don’t sound that great to me. And the PDF describing them is now apparently behind a paywall…
There is also some discussion about them in this topic…
Positional variation matters, perception matters (a lot), but ear shape also makes a big difference to the upper-mid and high peaks and dips. That brain is probably normalizing a lot of that. But why expect the immediate result of +5 to be perceived +5?
I have used Owliophile a few times. It is a wonderful idea - merging :
Which you then have to copy to your own EQ.
I found it more efficient to use an online tone generator (link below) for the tone generation, while I created the correction filters, based on listening, right there in my preferred EQ tool. This way, I have nothing to copy, after the correction is done.
And I can then,
1. Modify the baseline set of correction filters to create variations. The EQ in software which I’m using - Izotope Ozone EQ has a global amount parameter which defaults to 100%, but by varying this below and above 100%, one can adjust how much of the EQ is applied, to amplify or reduce the correction. Which I think is a wonderful feature or use a 2nd layer of EQ, for example to
Add additional personal preferences, such as a “room FR tilt” emulation.
AND/OR
Roll off (or boost) highest and lowest frequencies, to personal taste - this is important, cos listening to a tone generator, these frequencies are the most attenuated in our hearing, cos of Fletcher Munsen, so its better to tweak these with real full frequency range sources such as music. Not ideal to tweak these frequencies via listening to a tone gen.
Yeah, this is my preferred setup these days, with layered corrections where on one hand I’ve got the headphone-specific “neutralizer” as a constant, but on top of it I’m also running a manually tweakable EQ with 2 low-shelves and 1 high shelf as “tone controls” (down bot’ I use a sub-bass and an all-bass, not always both boosted - sometimes I like to cut sub-bass even while boosting all-bass). This allows me to fix various wacky mastering choices as Shuffle jumps me between recordings from different genres an eras.
Anyone else find it frustrating that you develop an EQ profile by ear that’s so good you want to share - but can’t! 
