Definitely. I look at graphs uh… way too often, maybe, and the “google image search” for, i.e., “HD58X graph” will show 10 different graphs… and half will be inaccurate for your ears.
For example, I tend to trust rtings because their measurements are what I hear. But you may have different ears… and prefer Inner Fidelity graphs, for example.
…And I don’t trust massdrop graphs.
Again, to each their own.
[Not sure how this got marked as a reply to M0N, I made it as a reply to MazeFrame.]
Very nice illustration! Sure hope this gets noticed and put into place. Something has to be done.
Here are the raw measurements of three different headphones each done by the three different technicians:
– Lies, damn lies and FR graphs, part 2 –
There’s enough variance after 1 kHz to sink two battleships plus an aircraft carrier. A competent rig used by a competent technician should agree within the limits of unit-to-unit product variation.
The conventional wisdom is that this means we can’t compare measurements between different measurement rigs+technician combinations. That’s enough of a pain. But I’m not seeing any evidence that any one of them is reliable from one headphone model to the next.
I can understand that going from the old straight-tube ear canal to a anatomically correct ear canal would cause issues. But even identical equipment seems to produce significant differences in results. Until this is ironed out FR graphs are no more reliable than subjective verbal reporting so far as I can see.
Here’s a practical example. I have a DT 1990. For EQ the Oratory measurements are the only ones that work, but they work very well. I also have an M50x. Not one of the three FR graphs produces anything like an accurate EQ result. I’m sure pad wear explains part of the difference – but surely only part. To get the M50x to match the 1990’s tuning I have to tune the M50x entirely by ear.
IAC, we all seem to be on the same page. I’m just adding yet more examples.
Really the best way if you cant hear it is to listen to a review. Zeos is good because he seems to like a lot of different types of sound from headphones and gear. And he has a LOT of experience cause of all the different headphones and gear hes heard. So i’d consider him a more neutral source.
Although he loves the Nueman NDH20’s. I dont understand how anyone could love a headphone like that lol. So its about taste also. One guy can love something i hate. So its best to get as many opinions as i can to get a better picture. The frequency graph can be a big help in this. Thanks to this post i can read it a little better now and im a little more wary of them too.
I dont listen to the sound tests Zeos posts. Wont they be skewed by the headset i listen to them on? How do i listen to them. Can anything be gleaned from them?
I dont think theres any other reviewer on youtube that as accomplished as Zeos and he seems to have more reviews on you tube than anyone else. So what other reviewers are as accomplished as him? what are the reviewers you guys listen too?
They are pretty fundamentally flawed tbh. It’s just not worthwhile imo as there are so many things that affect the final sound that it really doesn’t tell anything
I feel like there a decent amount of reviewers that do a good job at providing their perspective so I mean if you can find someone who has similar tastes and experience it can be helpful. Just watch out for obvious red flags. There is a thread on here somewhere about good reviews and people to stay away from or something similar that might be helpful to you
id really like to read that thread. Can someone link it?
Have you tried the NDH20s?
i bought them yeah. ugh, omg, 
.
terrible, just terrible
ah, they are my favorite after the Elex.
doesnt that muffled sound bother you? their so dark
I personally wouldn’t call them dark,they sound pretty reasonably pretty bright. LCD2C and Senn 650/600 are darker IMO. They have some of the best timbre and natural sound with instrumental music I’ve heard.
Well this is hilarious. Occurred to me to check what graphs are available for the Aiva to see whether we could glean anything more about it. So I did a google search with this result:
Three copies of the same graph you posted but none of them attributed to the actual source, which is the Aiva review by Jonathan Parks AKA Resolve Reviews. But the fourth graph looked truly bizarre to me. So clicked on it – and was taken to a post I had made in response to you regarding the Aivas back when you bought them! (The Sendy Aivas are fantastic!)
That graph is a screen grab of Oratory1990’s measurements of the Orpheus HE-1 that I used as an example of a point I was making in my comment. So anyone who isn’t careful or doesn’t understand FR graphs might mistake the HE-1 graph for an Aiva graph, lol! (I’ve got to remember to label all graphs I post right on the image itself.)
Getting back to the Aiva’s graph. Best guess is that is was compensated with MiniDSP’s HPN compensation curve, which in turn is apparently their version of the diffuse field curve. The DF curve traditionally looks like this:
IOW, a flat bass, gently rising mids but a massive upper mids peak topping out at roughly 15.5 dB. Due to a misunderstanding fostered by Sennheiser and probably others, DF is not a realistic version of accuracy but a large segment of the headphone community is used to graphs based on it. IAC, if we translate from DF to Harman or something along those lines, we have to add between 3 and 5 dB to a DF compensated graph at 3 kHz and taper off both before and after that. The upshot is that Resolve’s graph of the Aiva seems to show that it is recessed throughout most of the upper mids and treble. But in fact it may well show as only being recessed between 2 and 3 kHz if the graph were compensated by a Harman-like target.
It looks like all of them tested entirely different headphones, or mixed up graphs or something.
I’m posting 'way too much in the thread I started, but I just can’t help responding to this:
What a perfect example of graphs complementing verbal descriptions and vice versa. Here’s the Oratory1990 measurement of the Neumanns:
(It’s only the graph in the upper left corner that is relevant. Everything else has to do with EQ. Let’s see if Google manages to scrape a linked graph out of this thread rather than a pasted one…)
To me this graph says the headphone has an emphasized bass that extends a bit too far into the mids (to 400 Hz instead of 250) creating a bit of the very common mud region bloat. Then it fails to rise toward the upper mids and is thus noticeably recessed between 1 and 3 kHz. After that it looks to have an exceptionally nice treble.
Assuming the graph matches the sound of the pair of Neumanns both of you have/had, then the graph pretty much supports both your perceptions (big assumption!). Technically, Ant probably shouldn’t have used the word dark (treble roll-off) as an equivalent to muffled, but the combination of a bit of mud plus upper mids recession certainly warrants being described as muffled in my book. Yet there’s that beautiful treble that Giova is reporting.
Frankly, I suspect it comes down to both of you tending to listen to different music. Some music will have little content that is noticeably affected by the mud+recess and other music might well red-flag it to anyone’s ears. The DT 1990s I use for music listening has pretty much the same issues. So one piece of music will sound at least tolerable but the next will make me cringe.
Honestly google images is just like that. Search for an actor/actress and you’ll get pictures of his/her friend too, etc. You have to click and look at the source.
If someone has two graphs for two different headphones on one page, there’s a 50% chance google will show you the wrong graph, or a graph with pads swapped, or a graph with mods. I mean, search for “T50RP graphs” on google images. Good luck.
Yes this a million times this.
I find recordings I find poor on one headphone/Amp combo, will sometimes come alive on another.
If we can get away from the flat line response is good mentality.
Really where all of this measurement needs to start is understanding what people perceive to be “good”, is there even consistency there?
Clearly people have different preferences, but can we classify people into a smaller number of buckets to make recommendations easier?
Yes. A flat response is only good for four things so far as I can think. A) If you want accurate timbre, spatial placement, etc. when listening to acoustic music. B) If you want a good tuning compromise because you listen to a wide variety of genres on the same headphone. C) A hypothetical reference point for ease of comparing headphones. D) Certain aspects of recording studio work.
Variation in outer ear shape, ear canal shape and length, genre preferences, recency bias, and many other factors ensure there is no one-size-fits-all sonic goodness.
I think that’s what Zeos and DMS do with the headphones portion of their HiFi Guides purchase recommendation app. They divide sound signatures into dark, mid-forward, neutral, v-shaped, bright, and warm. Of course this presumes a degree of audio self-knowledge that the vast majority of consumers will draw a blank on. And veteran headphone enthusiasts will have that sorted out. It’s those who are new to the hobby who will find these categories a bit unclear. Seems to me that the quickest way to get up to speed for most of us would be to experiment with EQ.
I don’t think you need a flat signature to achieve this tbh
Yes
Can’t see how it could be otherwise, at least for timbre. Timbre is produced by the relative strengths of upper partials AKA overtones. Let’s say the fundamental is A 440. If there is a dip or rise that covers 880 Hz, for example, then timbre will be affected.
As for spatial placement, I’m just going off theory, since my hearing damage messes up my ability to hear this. But if there’s a dip in the mids then any instrument or voice sounding in that range would sound quieter than instruments or voices sounding outside that dip. I’d think something that is quieter than expected would be translated by the brain as being further away. And of course if louder than expected, the opposite.
That is somewhat correct, but I assume we can agree studio flat (flat line) does not sound realistically neutral? This is just me going off of my experience. For example, if I take the focal clear and eq it to studio flat, the timbre becomes less accurate imo. Stock it sounds better. I think there are some more things that goes into timbre, but for tone I would somewhat agree with what you say.
