Well, it achieves it in a different manner I suppose, if I understand the suggestion correctly at least. This is actually a good method of testing coupling consistency for both channels (it’s what Jude does I believe, although I can’t remember if it’s pink noise or white noise…) because it allows for a real-time view of the FR while you’re positioning it. I’ve also had to use it a few times to get FR for certain BT headphones. At the moment I find the sine sweep method faster/easier. Alternatively, what Crin and I will sometimes do to test coupling is do extremely fast or low res sweeps to make sure the seating is correct. But yeah if we ever get a 5128 we’ll likely do the Jude method of coupling consistency as well.
2 Likes
Intent of both test was to try and confirm that a simple Frequency Sweep is enough to capture subtle behavior of the driver.
If the white noise FFT closely (and it would have to be VERY close) matches the measure FR a of a sweep, then the sweep is probably sufficient.
Similarly if there is no change based on output level, and this would have to go all the way down to nearly inaudible (because isn’t that what detail is), then a single volume sweep is probably sufficient.
I’d personally want to include supersonic frequencies in any sweep/measurement as well because recent work on hearing suggests inaudible != inperceptible.
And this is before we get into the over simplification of testing or reviewing components in isolation, when they are never used that way, synergy is a very real thing, and overlooked once you start concentrating too closely on measurements.
I’m very much not against measurements, they are an interesting exercise, and I think it would be interesting to try training ML models on what we believe is “sufficient” to see if there is a signal there, and how much.
At the end of the day it’s about what you as a consumer hear, and what you like.
My top 3 headphones in order of preference FWIW Susvara, MySphere 3.1 and 1266 Phi TC.
But that ranking changes significantly if I can only listen to them on a DNA Stratus, when the Susvara and 1266 don’t even make the top 3. And there are other amplifiers I could pick where the MySphere wouldn’t make my top 10.
2 Likes
Yeah there’s definitely still a lot to figure out when it comes to this stuff, plus the more philosophical questions that are solely related to aesthetic judgements (not visual aesthetics… just aesthetics in general). Like if you just think about perceptual experiences and judgments about the way your music sounds, that’s a much bigger topic than what sound does in given conditions.
I’m interested in this as well, but largely because I’m curious if there are unique properties of transducers in the audible band that also have the consequence of being able to move fast enough to hit 40khz for example. Like if the subjective qualia could be different between two drivers where one can go to 40khz and the other can only go to 20khz. In both cases they reach above the bounds of human hearing, and theoretically there should be no difference. But what if there is… Tough to devise such a test though because you’d have to make sure that everything else for the FR is exactly the same. Otherwise any differences would be attributable to something else.
4 Likes
I’ve derived frequency response from different stimuli before (swept sine/Farina method, white noise, and musical signals) - the results of one such test were posted on Reddit by a friend of mine. With two-channel FFTs, it’s trivial to extract linear transfer function from any signal you want - the reason we use the Farina method for headphone and speaker frequency response is that we don’t get any additional information about frequency response this way.
Bear in mind that this can be viewed through the lens of a transfer function - if there’s nonlinearity in output as a function of input, we get distortions. If that nonlinearity was present at the lower levels but not the higher levels, you’d see something like zero-crossing distortion in class-B amplifiers - which we don’t, in headphones.
If you’re referring to the semi-recent meta-analysis of “high-res” audio, there were methodological problems with some of the studies it includes…
9 Likes
This also a good point and something that shouldn’t be discounted just because it doesn’t show up in measurements. And I’d count comfort in this as well.
The whole point here is to create an enjoyable experience, and if some of that is how something looks or feels, it’s still a part of the experience.
I think the best example of this is physical media, while there is some argument to be made for high quality physical transports, a lot of the listening experience with physical media is just different, there is a ritual involved in selecting a disk, removing it from it’s case/sleeve and putting into the player. And because of that ritual there is a tendency to listen “differently” (often more actively).
2 Likes
No I was referring to “recent” studies in hearing generally there is a lot of medical research around things like why even though a persons hearing measures extremely well (as mine does), why they find it increasingly harder to isolate conversations in the presence of noise.
A lot of it seems to indicate a much more complicated heating model than bones and hair in the ear resonating with the sound, though that’s clearly a lot of it.
Some of that is suggesting that perception of phase, or specifically the initial impact of a wave is significant in overall audio perception.
Perhaps this is obvious given some peoples sensitivity to absolute phase.
Are you referencing “hidden” hearing loss here? I’m not aware of anything there that challenges the hearing model - bear in mind, a lot of the “hidden” stuff is brain-ward of the cochlea, which isn’t contesting the acquisition process for sound, but rather clarifying how our brains process it (and can go wrong in doing so).
Bear in mind, in headphones - and all minimum phase systems - phase and magnitude frequency response are directly corelated.
In loudspeaker-room systems, there’s more significant bugbears associated with phase, which is part of why Floyd Toole’s wonderful book Sound Reproduction contains a section on it, part of which is pictured below:
Happily for those who are very worried about “correct” phase behavior, a headphone whose frequency frequency response is ideal and which does not display excess phase will also have the ideal phase response, and normal minimum phase IIR EQ will fix both at once
9 Likes
Things just got real in here. Lol.
7 Likes
Admittedly I fall into the “EQ profile” side of things, at least at this point in my life. I’m curious as to why you do not recommend liberal use of EQ, especially considering how much more information you think there is in FR besides overall tonality.
I’m still in the process of organizing my thoughts on “the case against EQ” video I want to make, even though of course I use it all the time haha.
In some ways it’s because there’s so much more information in FR besides target adherence/deviation that EQ has the potential to screw things up. The more you do it the more you realize the limitations of EQ.
I think oftentimes the argument against EQ is something along the lines of it ruining technicalities or limiting ‘dynamics’ and so on, and I’m not going to rule that stuff out here because there are times when I hear something with liberal EQ and I think to myself maybe that’s happened to one degree or another. There’s also the question of volume and pre-gain, but we do our best to hedge against that being as influential. But I also think there’s more to it than that.
There are more measurable and tangible reasons to not use EQ liberally, and it has to do with frequency response. While there’s valid argument that it’s destructive to the more fine-grained elements of FR (something that’s important to explore), there’s also the issue of unit variation causing massive problems for EQ profiles. Like if a resonance is shifted to a different spot, the profile can potentially fix it on one unit and make it worse on another.
The other aspect to this - and perhaps even more noteworthy - is that we’re not looking at the resultant FR at the ear drums of individual people (nor considering their HRTF, but that’s going to be an issue regardless). This means we can’t be sure of the way those changes are affecting the FR with our unique coupling situation and ear gain, even if we can be sure of it on measurement rigs.
Additionally, I think it’s important to consider that the targets we use are smoothed to 1/3rd or 1/2, meaning they aren’t ‘high res’ enough to perfectly indicate the various ear and canal resonances we have. So if you just outright match the coarse grained or highly smoothed target, it’s kind of like doing surgery with a blunt instrument.
Like if you take a headphone that has a ragged treble response between 5 - 10khz with all kinds of peaks and dips (which would be less desirable than a smoother response), but the overall level lines up reasonably well with Harman, if you smooth that to the same degree as the target it’ll look like it matches. This means that we really don’t know what the ‘high res’ version of that should be - we don’t know what to match to a sufficiently precise degree (consider that the 9khz concha notch is desirable but the target doesn’t reflect that for example).
What’s more, even if we did use an unsmoothed version of the target with the anthropometric pinna on a GRAS, we still wouldn’t know what that is at the ear drums of individuals. And… maybe the target has to be smoothed in order for it to be effective because of the varying gain factors across a wide range of people.
Of course, there’s also the last point that not all headphones are going to respond equally well to EQ, and yeah there are THD reasons to be careful about that.
If I remember Oratory’s comments on this stuff though, it was something along the lines of EQ being a crutch, and also undesirable, but perhaps the lesser of two evils when you consider some of the strange tunings out there. While I agree with the sentiment, I think my threshold will be a bit looser on that because the brain also does quite a bit of work for us. So for me it’s more about A) fixing major issues, and B) wide adjustments for preferences, like adding a bass shelf for example. For the rest of it, I’m fine with that just being ‘headphone character’, within reason of course haha.
So yeah, my favorite headphones are the ones I don’t really have to EQ much, or at all. But oftentimes - like the ones at the top of my ranking - the adjustments are very minor.
10 Likes
Your points are very thoughtful and now I find I have even more questions (and a few counterpoints). I could discuss this with you for quite some time but I don’t think this is the place for that.
One thing I will say is that I don’t think using more bands of parametric EQ is necessarily worse than using fewer, as long as the Q values are low enough.
You’ve given me a lot to think about. Thanks. 
1 Like
The only “defensible” line for review purposes may be not to use any post processing (including no EQ)? Which you have been doing as a baseline - I vote for a continuation of that approach.
Otherwise there are so many cool plug-ins that change the sound so much that it becomes impossible to compare headphones. E.g. I am running (this morning) Gullfoss Master, Black Box HG-2MS, Midside, Can Opener Studio, EQuilibrium – the difference to the “unfiltered” sound is quite big …
Yeah of course, especially the measurements have to be done without any additional effects, including EQ.
I agree with everything you’ve said here. I used to create the full EQ profiles and they always ended up sounding weird or blunt or just not quite right to me. I’ve transitioned to much simpler EQ now.
For instance, on the Focal Clear MG, I’ve added a 4db bass shelf at 100 hz, -4db at 1300 hz which kills the honky mid forwardness, and 3db treble shelf above 3k to lift its slight darkness. They sound really great to me now. With the Arya, I boost bass 4db and kill treble 2.5db (only on some recordings) and they sound just right.
I like high and low shelves because they change the tonal balance without messing with the character of the headphone to much.
3 Likes
That’s what it comes down to for me. A lot (maybe most) of the character of a headphone is in its frequency response. If I bludgeon it into following some predetermined target, it always loses that character. Targeted adjustments can maintain what I like while fixing what I don’t.
1 Like
Ideally the only thing we would do for open-back headphones is add a bass shelf. That’s the dream.
Reason being, we shouldn’t really expect open-back headphones to perform ideally in the bass with exactly the bass shelf the target asks for because as I’m learning, there are acoustic challenges to that in the first place. To me this indicates two things, one is that it doesn’t make sense to take marks away from an open-back headphone for not achieving the target bass shelf, the other is that we might as well EQ that in anyway if we desire it and the headphone in question can handle it. In my view, there’s really no downside to that type of adjustment.
8 Likes
It is definitely an interesting discussion and debate over EQ. I know since getting into this hobby (headphones that is) I personally have not EQ’d anything. Since reading everything through here and watching @Resolve reviews I finally figured it out for myself. It’s not laziness…hell if I didn’t want to mess with software there is a very nice thing called the Lokius, which I admittingly have been eyeing.
Growing up I never hesitated to crank up the bass a bit more when I had stereo systems in my car. After all, that’s just what we did…or at least I did when growing up with my friends. I’m talking in the 90’s to date myself. I never considered that EQ - and yes I know it is.
That’s also because back when I was buying nice sound equipment I never considered sound quality as in Bass vs Treble (mids too). To me it was: You get a sub for the bass and let the other speakers be your mid/treble fill. Up the bass a bit for some more oomph and you’re good.
Back to headphones - there is no sub, no tweeters…just the drivers. So I research for what I want to buy based on sound signature and then feel cheap if I tweak that sound with EQ. Now - I’m sure that is what this whole thing is about. How you shouldn’t feel that way about a product because it’s your money, so cater it to yourself.
Even though I admit that’s the main problem is a senseless sense of pride for the company who made the device…it is what it is. I 100% appreciate what you do for this hobby and enthusiasts. It get’s me to think and look at things like this differently.
Anyway, my 2 nuggets here don’t really add any value to this conversation other then just to comment. I like the tier list of EQ and non-EQ. I think it’s needed for any kind of tier list when it comes to audio. Hell even how certain amps can subtly change a sig too. So a big thank you is needed!
1 Like
This is exactly the option I’m starting to explore. I got into EQ through applying Audyssey EQ to my home theater system. Now the home theater experts are suggesting that for best results Audyssey and similar EQ solutions should only be applied below 200-300hz if you are using good speakers to begin with.
Amen! Just a touch of EQ… less really IS more in this case.
1 Like
Thanks for the hard work, the list is very helpful. Would like to see your list for IEM, especially with the Monarch mk2.