• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

The science behind Stax's magic

Ata

Senior Member
Forum Donor
Joined
May 6, 2021
Messages
320
Likes
259
Location
Adelaide, Australia
Moondrop Kanas Pro's (sadly discontinued, but their sucessors like the Starfields would also suffice) have vanishingly low THD, to where I've tried them unsealed just hanging loose in my ear canal, and I can't hear distortion (if I had them sealed, it would literally hurt for more than a split second of sound).

You also have Sennheiser IE 40 for $98 on Amazon Prime shipped, which are almost identical in terms of THD spec (they also sound like their headphones somewhat tonally speaking). Amir also reviewed some Panasonic IEM (the RP-TCM125) which had insanely good THD

I have ordered the Panasonic IEM for the asking sum of 11 USD and can experience first hand how it sounds after the insanely good (on paper) EQ.

But, why stop at THD metrics, what about other forms of distortion: IMD, resonances, compression, etc? Seems like one should focus on the other distortion metrics as well...
 

Tks

Major Contributor
Joined
Apr 1, 2019
Messages
2,936
Likes
4,640
I have ordered the Panasonic IEM for the asking sum of 11 USD and can experience first hand how it sounds after the insanely good (on paper) EQ.

But, why stop at THD metrics, what about other forms of distortion: IMD, resonances, compression, etc? Seems like one should focus on the other distortion metrics as well...

Sure, but I wouldnt know how to measure IEM resonance for example. I also am not aware of what compressions actually are as a distortion metric. THD sinplifies this as it takes the TOTAL of all distortions (as ive never seen something like IMD be higher than a general TDH summation to where I woupd only hear IMD while not hearing any others). IEM distortion is so low to begin with, I have trouble hearing any (and sometimes mistake it for the music itself and it's poor recording sometimes).

In general I agree with you. Practically, I just simply have no clue if this is something that warrants particular concern.
 

John_M

Member
Forum Donor
Joined
Jun 17, 2020
Messages
61
Likes
34
Is this the 'received wisdom' of the forum?

It goes to the question of whether it's worth buying Stax. There are cheaper headphones than (say) the 009 which have similarly low distortion levels.

Or, to take a more extreme example, the HE-1. If this were right then the HE-1 would look like a colossal waste of money.

Nobody's answered this question. :)

Is it considered a waste of money to do anything other than buy a very low THD headphone and EQ it? (Assuming it's comfortable/ well built and there's nothing else materially wrong with it).

I know there are points about HRTF which mean that EQ'd headphones may not sound identical to each other but this doesn't seem to mean that the EQ'd headphones would necessarily sound worse.

Or, putting it in the terms of the thread title, is there some other undefined "magic" that (say) Stax has?
 

Shazb0t

Addicted to Fun and Learning
Joined
May 1, 2018
Messages
518
Likes
849
Location
NJ
Or, putting it in the terms of the thread title, is there some other undefined "magic" that (say) Stax has?
There is no magic. They pressurize air in waves from a voltage input like every other headphone.
 

Ata

Senior Member
Forum Donor
Joined
May 6, 2021
Messages
320
Likes
259
Location
Adelaide, Australia
Nobody's answered this question. :)

Is it considered a waste of money to do anything other than buy a very low THD headphone and EQ it? (Assuming it's comfortable/ well built and there's nothing else materially wrong with it).

If you read my post above you will see that I do not believe linear (FR) and non-linear (THD) distortion is all that one has to care about. There are transparency and soundstage that also play a part in the subjective evaluation of sound transducers, and it appears science has not yet caught up on how to measure these.

I personally believe IMD plays a big part when any complex music is being played. Have you heard speakers and headphones/IEMs that sound really nice with "simple" music (piano, guitar, vocals) but "fall apart" as soon as more complex music (classical, rock) with multiple instruments "filling" the audible spectrum with energy? My completely unscientific theory is this is due to IMD, and that this is why well implemented multi-driver IEMs and speakers generally do better at complex music than similar transducers with fewer drivers, with better "separation" between instruments and "more air". IMD is then limited per transducer and cannot "leak" between transducers.

However, IMD measurements are apparently difficult to do, and non-standardised. They are also not well understood. Some sources claim IMD measured at the transducer can vary highly with the choice of amplifier design. Not only that, but they also measure a tube amp to provide lowest IMD, and a Class D amp highest:

https://josephcrowe.com/blogs/news/...-choice-on-speaker-intermodulation-distortion

I have no idea how reliable this information is.

I know there are points about HRTF which mean that EQ'd headphones may not sound identical to each other but this doesn't seem to mean that the EQ'd headphones would necessarily sound worse.

Correct -- you can take a pair of open cans and move your hands close to the cups while listening and see the effect of HRTF very clearly. You can make them sound "better" that way, "open up" the soundstage, etc.

Or, putting it in the terms of the thread title, is there some other undefined "magic" that (say) Stax has?

There is no "magic", but one theory is that electrostatic transducers, due to their extremely low weight, are less adept at generating bad THD (odd harmonic distortion) and less IMD?
 

MayaTlab

Senior Member
Joined
Aug 15, 2020
Messages
401
Likes
560
I personally believe IMD plays a big part when any complex music is being played.
There is no "magic", but one theory is that electrostatic transducers, due to their extremely low weight, are less adept at generating bad THD (odd harmonic distortion) and less IMD?

IMD has been known for years already to be very poorly correlated with how listeners subjectively hear distortion :
Among many others : https://projekter.aau.dk/projekter/files/9852082/07gr1061_Thesis.pdf
As already mentioned, there are plenty of headphones with lower THD than Stax's, far below threshold of audibility regardless of the harmonics, and I wouldn't be surprised to learn that it's the same with IMD.

The likelihood that THD or IMD has anything to do with how people subjectively perceive Stax headphones is, IMO, very low, or at least negligible compared to FR at their own eardrum. In the former two cases we're dealing with metrics for which we're not even sure they're audible given their lower prevalence with current headphones, while for the latter we're dealing with variance between listeners that we know significantly exceeds threshold of audibility.
 

Ata

Senior Member
Forum Donor
Joined
May 6, 2021
Messages
320
Likes
259
Location
Adelaide, Australia
Electrostats have very low THD to begin with, in the 0.1-0.2% range based on the few measurements I have seen. That level is likely inaudible.

I just took delivery of the Panasonic IEMs recently reviewed here, those which have comparably low THD and can be EQed to unreal levels in terms of metrics. I will know if the 11 USD earphone will sounds similar to the 400 USD Stax in no time :) and report back my subjective experiences.
 

MayaTlab

Senior Member
Joined
Aug 15, 2020
Messages
401
Likes
560
I just took delivery of the Panasonic IEMs recently reviewed here, those which have comparably low THD and can be EQed to unreal levels in terms of metrics. I will know if the 11 USD earphone will sounds similar to the 400 USD Stax in no time :) and report back my subjective experiences.

Well, you probably can't measure their effective FR at your own eardrum, so we already know that you won't be able to EQ them to reach the exact same FR at your own eardrum as your Stax. And since that variable will remain uncontrolled, you won't be able to reach conclusions in regards to variables other than FR. That's the big problem we're all facing :D.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
10,615
Likes
23,510
Location
The Neitherlands
Add to that seal and insertion depth as well as positioning and seal of the Stax can be of major influence.
Fortunately our hearing is not as discriminate as one might think.
 

Ata

Senior Member
Forum Donor
Joined
May 6, 2021
Messages
320
Likes
259
Location
Adelaide, Australia
Well, you probably can't measure their effective FR at your own eardrum, so we already know that you won't be able to EQ them to reach the exact same FR at your own eardrum as your Stax. And since that variable will remain uncontrolled, you won't be able to reach conclusions in regards to variables other than FR. That's the big problem we're all facing :D.

While I agree in the literal sense, I am uncomfortable with this statement as it implies that "you can never compare or improve on your transducers, as you cannot measure them at your eardrum". There is a line one has to draw as to where the diminishing returns are, and to me I choose to ignore the inaccuracy of the measured/predicted vs actual as far as ear-to-ear differences are concerned.

Things become messier and fuzzier once you start looking at the FR of your own eardrum and nervous system / brain, and its function of the volume you are listening at.
 

MayaTlab

Senior Member
Joined
Aug 15, 2020
Messages
401
Likes
560
I choose to ignore the inaccuracy of the measured/predicted vs actual as far as ear-to-ear differences are concerned.

This is the one thing you absolutely shouldn't ignore with headphones (I can't emphasise this enough). If only for sealing issues or insertion depth, as Solderdude just mentioned. This sort of variation is far, far above threshold of audibility and way, way more important than any concerns regarding IMD or THD :
https://www.rtings.com/headphones/1-3-1/graph#669/3185

In reality the response at a particular listener's eardrum across the FR spectrum can be evaluated to a degree, but more at the design / engineering stage. One way to do it between around 200hz and 10 000hz is to use probe tube microphones, like this : https://www.etymotic.com/auditory-research/microphones/er-7c.html
But they require rigorous placement to get meaningful results and only a few studies I've read managed to coax sensible looking FR curves out of them. I like this one in particular : https://www.aes.org/e-lib/browse.cfm?elib=16877
Below 1khz, for over-ears, you can use in ear mics to get pretty accurate values, such as these : https://www.soundprofessionals.com/cgi-bin/gold/item/MS-TFB-2. This uses the same principle ANC over-ears already use to measure FR in real time below 1khz. Obviously not applicable for in-ears.
Manufacturers of ANC TWS with inward facing mics (AirPods Pro for example) can use them to measure response below a few kHz down to subbass levels.
So if HP manufacturers actually care (I doubt that all do), they're not working "blind", but other than the in-ear mics for over-ears, amateurs such as you and I have only limited means to do the same.
 
OP
Yoaime

Yoaime

Member
Forum Donor
Joined
Apr 16, 2021
Messages
13
Likes
3
Location
Belgium
I got to test some Stax (202, L300 & L500) today, spent around 2h30 with them. There was also a HD660.

Started with the HD660 that sounded tight/precise compared to my M1060. In comparison the M1060 sounded dull, soft. They weren't much different for everything else I would say.

Then I put on the Stax from cheapest to most expensive. They are definitely more similar than different.
If we take the 202 as base, the 300 has a bit less low-end but sound more crisp in the highs and is a bit sibilant.
The 500 remove that harsh "s" from the 300 and add deeper and more present low-end.
But that's really minor. That, I'm sure of.

After that, I don't really know. They sounded different because the tonality is and the soundstage seems a bit wider and some other things maybe. Not better or magical but different.
Stax don't just have less bass and that's it. The L500 definitely gives you bass. I think they just have more forward mids and low highs, combined with peaks in the highs. But after some times my ears were too tired and all headphones sounded the same so I couldn't check anything anymore. Music stores aren't exactly quiet (especially this one) so I had to turn the volume up and got tired quickly.


While switching between the 660s, M1060 and l500 sometimes I was using one with the EQ preset of another for a few seconds and it sounded like the headphones I was putting on, weird but still recognizable.
While at home, for weeks now, I tried to EQ my headphones to sound like each other. Or more like trying to get the best of each one to the others to end up with my "perfect sound". My SRH1840 was hopeless, at least I couldn't find a way to get rid of what makes me think this one sound bad. But I could do something for my Sony XM4.
Now they sound really near my EQed M1060. The M1060 are a little more open sounding with a bit wider soundstage and do much better 20-70Hz, for everything else they are pretty damn almost the same. After 30 seconds of listening I forget which one I'm wearing the difference of 'openess' and 'soundstageness' fade away.

But it's nowhere as simple as slapping Oratory1990's preset to match Harman curve on both. And that's where it gets weird. I had to spend hours and hours to find the right EQ manually to match the M1060. With the preset measured I got a bit closer, then I worked on that and applied another heavy EQ on top.
- Do I compensate for other factor than FR with my EQ ? Don't think so
- Oratory measurement are wrong ? They can't be that way off anyway.
- My head/ear shape react differently so I have to compensate for that. Maybe, but they both have to go through the same ear canal, so why would they be different ? My ear can recognize the headphone I'm wearing and change shape accordingly ?
I know that anything above 8k cannot be measured properly to cannot be EQ this way, but my EQ affect the whole spectrum.

My current belief is that you can take any headphones that have low enough distortion (and no weird ringing), can do the deep bass and soundstage as you like, then EQ it to whatever sound you like. But it takes hours and you have to have the desired target at your disposal at home (like the band in you living room for the best result) for comparison the whole process.
But my beliefs are shaken pretty often in light of new data.
So my take on the previous posts is that transparency would just be FR and soundstaging capabilities.


I surely have missed something. I'm considering buying a Stax setup to test my theory.
Should have picked audio engineering instead of informatics. :)
Sorry to dig up this thread. I think it is worth to share my new findings.

So basically I was wrong.

At the store I was impressed by the dynamics of the HD660, and I couldn't get it with EQ on the M1060. A bit of exageration on 200Hz helped but was nowhere near the HD660. I also remembered my good impressions while quickly test the Focal Clear at a store a couple years back.
I wanted to test the Elex, but it's not sold in Europe and having a closed pair is nice live near a busy street in summer. And I was pretty sure my physical pain when listening to music was caused by the M1060 doing something wrong (which it was).
Anyway I got myself the Focal Elegia. Blindly bought over the internet, trusting the reviews.

God this sounds good !
Dynamic, imaging, details, transparency, clarity... whatever you want except soundstage is greatly better than M1060.
The Elegia allows me to hear things more clearly, to continue to hear the bass line even when the guitar and singer get excited.
It's not because of frequency response neither the THD. If I alter the FR on the M1060 to better hear something, I lose something else.
So while we wait for good ways to measure those characteristics, we have to trust detailed subjective listening experiences if we can't test the device.

Side note on testing in store : It doesn't works, the impressions are wrong because you come with your brain tuned to what you have at home. And over time the brain adapt itself to frequency response, that's impressive.

Happy listening to everyone :)
 

posvibes

Active Member
Joined
Jul 4, 2020
Messages
101
Likes
159
I've been listening to my very old Stax Lambda SR's lately with much joy. But only because I discovered how much better they sound with Peace EQ "Bass Boost 2 setting".

The setting has all sliders set to 0 gain and one slider with a lower shelf boost of 10db at 166Hz with a minus -30db preamp setting.

They just came alive.
 

Graph Feppar

Active Member
Joined
Nov 19, 2018
Messages
245
Likes
215
Stax magic is simple, ultra low distortion + flat(ish) and smooth FR + flat and smooth phase response + quick decay + planar wavefront + low excursion to spl ratio.

Stax with good seal have very low distortion all across the spectrum. Even with the stupid build in ports and with leaky pads the distortion is only increased in bass to level of average dynamic headphone. The low distortion is the key as it allows low level detail that sits barely above the theoretical perceptual threshold, or as SBAF call it, the "plankton", to be audible becose it isnt lost beneath the sea of intermodulation products like in normal headphones.

They have flat and smooth frequency response and phase response, they do this effortlesly, its inherent characteristic of this technology. The stators are stiff and pass sound easily so they ring less than planar magnetic stators or dynamic drivers and the diaphragms are unusual in their low tension. While planar magnetic headphones need high tension which causes it to ring like small 10 KHz bongo drum, electrostats only need minimal tension with resonant frequency down low around 100 Hz.

The diaphragm also weight very little as it doesnt need to conduct current unlike planars which causes air to strongly damp the fundamental resonance, that is why ported and leaky stats have sharp peak in bass followed by steep roll off while ones with good seal are super flat in bass.

This combination of low diaphragm weight and low tension prevents the diaphragm from having high frequency break up modes, a uneven resonance of the membrane caused by playing HF tones with wavelenght shorter than the size of the membrane.


The massive size of the diaphragm translates to planar wavefront, this makes the ear transfer function to be similiar to real world so it gives them good soundstage. The cherry on top is ratio of excursion to spl which gives them extra low phase intermodulation distortion. They dont need to move very far to move air becose of their large surface area.

To recap, electrostatic drivers are inherently more linear and less resonant than planars and dynamic drivers. Clean THD, clean FR, clean phase and clean CSD create the magic electrostatic sound becose the nonlinearities and resonances of normal headphones that normally hide the micro detail are absent.
 
Last edited:

2M2B

Active Member
Joined
Mar 9, 2021
Messages
118
Likes
70
For there to be this difference in DD vs BA, he would have to be able to AT LEAST blind test random IEM's where he doesn't know what the driver configuration is, and pick out which are DD and which are BA (since I've heard him go on record complaining about BA bass).
They didn't know the ER2XR used a 5.8mm DD when my last reply from them was that IEM's need 12mm when ranting about the ER4XR/SR being 5mm. They've refused to do a A/B by testing bass quantity like EQ'ing the ER4SR/ER2SE with the same bass as the 2XR. Then do a 2nd one with them back to flat bass on both since I've seen them claim the ER2SE is as bassy as the ER4XR which has 4db more bass than the 2SE.

I've got a ER3XR right now after trying the ER2SE with 4db boost like on the 4XR. I doubt I could tell both apart on slow to moderately fast music, But on stupidly fast stuff(grindcore, Noise rock, Speedcore, etc) the ER3XR holds up better.
 

Tugbars

New Member
Joined
Nov 17, 2021
Messages
3
Likes
3
Time domain is linked to frequency domain on min phase systems hence CSD doesn't provide any kind of useful information. FR & group delay give all the information we need. https://www.linkedin.com/pulse/csd-cumulative-spectral-decay-really-important-jason-dai/ a very good article about this.

Excess phase seems to be low on electrostatics compared to most other headphones. That means, calculated group delay from FR / measured group delay deviations are less hence most Stax headphones tend to have less phase shifts than most of their competitors. This said, Susvara's excess phase measurements are quite good too(Sundara also performs as good as Stax).

The resonance frequency of Omegas is around 50-60hz while the f0 of lambdas range from 70hz to 90hz. Hifiman planars also have low resonance frequency(60hz'ish) and Focal Utopia's resonance frequency is 50hz. I don't think Stax have an upper hand in that area.

I don't know any Stax headphones even with slightly good treble response. All of them have a nasty peaks throughout whole treble & high treble range. Omegas have usually a 1100hz peak(pads), a recession area around 2khz, non linearities around 5-9khz and a nasty 11-12khz peak follows them. I think Stax headphones generally are just horrible. They are shouty(1-2khz response have always some weird peaks & dips), 5-8k is either super recessed or super emphasized, 10khz+ is just out of control.

Low diaphragm mass is not always a good thing. The stiffness of the air surrounding the diaphragm is enough to acoustically damp super light diaphragms hence they can stop better: Yes, they have good treble extension. However, neutral headphones should roll off in high frequencies because neutrality(unless binaural recordings) is defined by the response of speakers in a typical room and well designed flat speakers do not have flat power response. Staxes have weird peaks around these areas instead of a roll off. This is completely far from neutrality & kill timbre of almost all instruments.

Distortion is rarely a problem on headphones because of how auditory masking works. Depending on the SPL and the frequency, fundamental tones tend to mask the second/third(or higher) harmonics. Sean Olive also mentioned in one of his interviews that during conducting blind tests & research for Harman, they couldn't find single pair of headphones with audible distortion problem. There is a really good AES article about the relation between auditory masking & headphones & distortion too. I can share the link if anyone wants. *Low level detail* also has something to do with how the peaks & dips interact with each other based on auditory masking principles. Usually different peaks and dips in the frequency response mask way more detail(frequencies around them) than distortion ever can.Here is a very simple explanation of how it works:

An-audio-masking-graph.pngg

The massive size of the diaphragm translates to planar wavefront, this makes the ear transfer function to be similiar to real world so it gives them good soundstage.
This seems interesting, I thought of this too but I couldn't find any research to back it up. Do you know any research & articles related to this phenomenon?

I have about 11 different electrostatic headphones(Omega 1-2-3, SR507, 404LE etc etc) I think all of them sound just horrible without EQ. I really enjoy them after EQing them to Harman 2013 target though.
 
Last edited:

Graph Feppar

Active Member
Joined
Nov 19, 2018
Messages
245
Likes
215
Time domain is linked to frequency domain on min phase systems hence CSD doesn't provide any kind of useful information. FR & group delay give all the information we need.
That is true, I wrote the same exact thing like week ago in one of my posts. We could go step further and say we dont even need frequency response graph, we just need impulse response.

Here is the thing, humans are not robots.
While impulse response technically contain all the frequency, phase and ringing/decay information, for us humans its practically impossible to obtain that information just by looking at impulse. I might see some high frequency resonance if there is long ringing with clean sinusoid cyclces but as far as phase and FR go, there is no chance I am getting anything out it.

Excess phase seems to be low on electrostatics compared to most other headphones. That means, calculated group delay from FR / measured group delay deviations are less hence most Stax headphones tend to have less phase shifts than most of their competitors. This said, Susvara's excess phase measurements are quite good too(Sundara also performs as good as Stax).
This is very interesting. For longest time, I thought real world mechanical system like speaker drivers, headphones or even music instruments cant have any excess phase. Where did you saw excess phase measurements of headphones? I never saw such thing in my entire life, would love to see some graphs if you have any.

Susvara maybe doesnt have much excess phase but it does have, like all other Hifimans, countless tiny resonances that audibly smear the sound into low fidelity mess. Hifimans are in my opinion most overhyped headphones ever, I cant understand why would anybody like them, on top their build quality is poor also and they arent cheap. Then I remember we live in world where number one headphone brand is Beats and it all makes sense.

The resonance frequency of Omegas is around 50-60hz while the f0 of lambdas range from 70hz to 90hz. Hifiman planars also have low resonance frequency(60hz'ish) and Focal Utopia's resonance frequency is 50hz. I don't think Stax have an upper hand in that area.

Hifiman planars definately dont have 60Hz resonace. They all have diaphragm tension such that the resonance is in the high frequency audible range which can be seen by massive ringing in CSD and gigantic distortion increase in same area.

Important fact about stat vs planar driver fundamental resonance is that in former, the resonance completly disappears with good seal while the later will resonate strongly no matter what you do.

Stax and other electrostatic headphones absolutely have upper hand in this area. Audeze did the right thing and burry that nasty resonance region in deep dip so its not offensive. Lowest tension planar I know is discontinued Oppo PM1/2 which had lowest planar resonance at just 350 Hz, they had the fastest decay and smoothest top end out of all planars by mile, go and figure.

I don't know any Stax headphones even with slightly good treble response. All of them have a nasty peaks throughout whole treble & high treble range. Omegas have usually a 1100hz peak(pads), a recession area around 2khz, non linearities around 5-9khz and a nasty 11-12khz peak follows them. I think Stax headphones generally are just horrible. They are shouty(1-2khz response have always some weird peaks & dips), 5-8k is either super recessed or super emphasized, 10khz+ is just out of control.
Ehh, ok, alright. I was going to write that Stax are renowed in audiophile community for their good treble but I shat on Hifimans which are popular too so I say we all get our own preferences and opinions, the beaty is in the eye of beholder as they say. I will say ( write) this, if Stax have bad treble, then what are Hifimans? An accoustic torture device? Ever heard about Ultrasone?

Dont get me wrong, I agree with all your critical observations about Stax highs, its just that after having climbed to the summit of mount Beyer and experiencing my hair cells being combed by comb filter like FR of Hifiman Edition X, I would personally not rate Stax so harshly considering what kind of headphones are out there.

Low diaphragm mass is not always a good thing. The stiffness of the air surrounding the diaphragm is enough to acoustically damp super light diaphragms hence they can stop better: Yes, they have good treble extension.
Low diaphragm mass is always a good thing. Air is gas, it has no stiffness, it has mass and compressibility. The lower the diaphragm weight, the better the damping, it rings less becose the air brake it more, and also efficiency as more energy is transfered to air.

However, neutral headphones should roll off in high frequencies because neutrality(unless binaural recordings) is defined by the response of speakers in a typical room and well designed flat speakers do not have flat power response. Staxes have weird peaks around these areas instead of a roll off. This is completely far from neutrality & kill timbre of almost all instruments.
Strongly disagree with this, Harman is not neutral, speakers in room arent neutral, rolled off highs are not neutral. Is that perhaps preferable to you and other people? Possibly yes, but is this neutral? Hell no! True neutral doesnt have boosted or rolled off anything.

I think Stax timbre is much more realistic with its extended top end and slight air area boost than other bandwith limited headphones or speakers that lack extension and they fall off cliff in treble making everything sound like low bitrate mp3.

Distortion is rarely a problem on headphones because of how auditory masking works. Depending on the SPL and the frequency, fundamental tones tend to mask the second/third(or higher) harmonics. Sean Olive also mentioned in one of his interviews that during conducting blind tests & research for Harman, they couldn't find single pair of headphones with audible distortion problem. There is a really good AES article about the relation between auditory masking & headphones & distortion too. I can share the link if anyone wants. *Low level detail* also has something to do with how the peaks & dips interact with each other based on auditory masking principles. Usually different peaks and dips in the frequency response mask way more detail(frequencies around them) than distortion ever can.Here is a very simple explanation of how it works:

View attachment 175266
Distortion is the biggest and most common problem on headphones. You can fix FR of any headphone with EQ, you cant do anything with distortion except improve seal.

The distortion masking works great when single instrument is playing but the moment another non integer tone appears ( drums, non autotunned vocals, detuned synths, transients, random noises ) , it all falls apart as entire spectrum gets flooded with nasty inharmonic intermodulation tones that lay far away from the frequency band of the music instrument hence arent hidden by frequency masking. Distortion is audible on all but select few headphones I tried and looking at measurements I say it will be easily audible on 99% of all headphone.

I dont like Sean Olive, I hate Harman curve and I dont care about his cool story regarding distortion as its laughably wrong from my own experience and research into nonlinear systems and human hearing.

I do agree on you that the small peaks and dips do strongly impact sound quality and hide detail. The most revealing headphones are always the ones with smoothest frequency reponse, that is why I love Sennheiser 600/650/660S.
 
Last edited:

2M2B

Active Member
Joined
Mar 9, 2021
Messages
118
Likes
70
The distortion masking works great when single instrument is playing but the moment another non integer tone appears ( drums, non autotunned vocals, detuned synths, transients, random noises ) , it all falls apart as entire spectrum gets flooded with nasty inharmonic intermodulation tones that lay far away from the frequency band of the music instrument hence arent hidden by frequency masking. Distortion is audible on all but select few headphones I tried and looking at measurements I say it will be easily audible on 99% of all headphone.
I'm going to trust Sean Olive on this since they actually use DBT data on large set of people. I love when people say this but go quiet when showed that there Speaker they use daily measures worse than anything Grado labs makes THD wise.

99% of all headphones are 0.4 ~ 1% at 1KHz+ but many OE's are 4% average for bass with planar/Estat being 0.5%. But any IEM will be under 1% full range, I highly doubt anyone could tell a ER4XR at 1% unless there playing pure sines at 95db?.

I've noticed you didn't answer my comment about the dudes on Head Fi saying that the ER4PT being 0.4% at 80db with a TOTL mic will be 0.4% at 100db.
 

Graph Feppar

Active Member
Joined
Nov 19, 2018
Messages
245
Likes
215
I'm going to trust Sean Olive on this since they actually use DBT data on large set of people. I love when people say this but go quiet when showed that there Speaker they use daily measures worse than anything Grado labs makes THD wise.
I dont have any speakers :( (semi suppressed sobbing)

Speakers that measure same THD as headphones will have much lower THD when playing wide bandwidth signals such as music becose speakers have multiple drivers. Speaker woofer might have 5% THD at 50Hz but due to the multi driver design, the distortion is contained in bass region while in headphones, it will intermodulate everything up to 20 KHz creating full spectrum IMD bukkake.

99% of all headphones are 0.4 ~ 1% at 1KHz+ but many OE's are 4% average for bass with planar/Estat being 0.5%. But any IEM will be under 1% full range, I highly doubt anyone could tell a ER4XR at 1% unless there playing pure sines at 95db?
THD at "1KHz+" doesnt matter much when highest amplitude signals in music are in bass region which also coincides to be the most distortion prone region in headphones. Many planars that have low bass distortion have sharp distortion peaks in mids and highs. There are very few full sized headphones that have under 1% distortion in entire spectrum at 100db.

As to your 99% number, what frequency and what SPL? From the measurements I saw, atleast 95% of headphones reach over 1% THD at some point in spectrum at 100db.

I've noticed you didn't answer my comment about the dudes on Head Fi saying that the ER4PT being 0.4% at 80db with a TOTL mic will be 0.4% at 100db.
Who are the dudes from Head Fi? I say they are noobs with crap microphones. Mistaking noise from your cheap measurement mic for THD of headphone is classic novice blunder.

No, the ER4PT, or any ER4/3, wont be 0.4% at 100db. More like 2% and that is in the 1KHz+ region.
e0f8355479844044b2e59a6c882f9d5c-0001.jpg
 

Tugbars

New Member
Joined
Nov 17, 2021
Messages
3
Likes
3
While impulse response technically contain all the frequency, phase and ringing/decay information, for us humans its practically impossible to obtain that information just by looking at impulse. I might see some high frequency resonance if there is long ringing with clean sinusoid cyclces but as far as phase and FR go, there is no chance I am getting anything out it.

The point is, for a minimum phase device(headphones) the resonances are inherently linked to the magnitude response. We can not hear the time aspect(ringing) of resonances.(There are researches about that. And yes, we are not robots. Yes, measurement rigs hear better than humans.). A min phase system can't have "ringing" without having a bump in the frequency response. And if you fix the bump in the FR, that would fix the ringing too. This is how it works.

The article I shared explains this in simple terms. I recommend you to check, It is written by engineer who works in Audio Precision.


Strongly disagree with this, Harman is not neutral, speakers in room arent neutral, rolled off highs are not neutral. Is that perhaps preferable to you and other people? Possibly yes, but is this neutral? Hell no! True neutral doesnt have boosted or rolled off anything.

I think Stax timbre is much more realistic with its extended top end and slight air area boost than other bandwith limited headphones or speakers that lack extension and they fall off cliff in treble making everything sound like low bitrate mp3.

Again... wrong. Super wrong.

Genelec 8361A Measurements Predicted in-room Frequency Response Powered Studio Monitor Speaker.png


Genelec 8361 has extremely flat on-axis response however its in room response is a constant downward slope from 20hz to 20khz. Why? Because low frequencies are omnidirectional while higher frequencies are more directional. Speakers can not emit high frequencies with the same acoustic output as low frequencies unless they are omnidirectional. As a result, power response of speakers drop in higher frequencies and listening window response + early reflections(and sound power) form such a downward slope from 20hz to 20khz. This is how the estimated in room response is calculated. These estimations done by Klippel NFS are extremely accurate after the room transition frequency. Speakers sound almost exactly how the Klippel system estimates after 300-500hz~ in rooms. I can share Erin's data about this topic if you want. (Room transition frequency is usually somewhere around 300-500hz) (note: measurements taken by omnidirectional pressure microphones)

The music we listen is recorded with speakers in typical reflective rooms. What kind of speakers do audio engineers use? Omnidirectional? No. Since headphones reproduce the information that is mixed/mastered with speakers, they are neutral as long as their response matches with the in room response of studio speakers. Headphones that matches this curve after the ear gain compensation can be considered as *neutral*. (assuming audio engineers use state of art speakers.).

The distortion masking works great when single instrument is playing but the moment another non integer tone appears ( drums, non autotunned vocals, detuned synths, transients, random noises ) , it all falls apart as entire spectrum gets flooded with nasty inharmonic intermodulation tones that lay far away from the frequency band of the music instrument hence arent hidden by frequency masking.

Again, wrong. The amplitude of masking threshold of differs with loudness of the fundamental and frequency. Therefore, usually second or third order distortion gets masked(unless the fundamental is very low) however 4th order, 5th order, 6th order distortion can escape from the masking threshold of the fundamental. Transducers rarely have such high order distortions. Sean Olive's data, Earl Geddes' data tell us a different story than what you are telling me, If I recall it right, during the blind tests, volunteers couldn't hear 20% distortion of a compression driver(yes they can go very loud).
A nice article about headphones distortion and its audibility: https://www.aes.org/e-lib/browse.cfm?elib=19238

At any rate, headphones do not suffer from IMD distortion. IMD distortion has something to do with excursion and/or bandwidth of the system. The diaphragm excursions on headphones is super low, if you have any article about the IMD distortion of headphones and its audibility, let me know. Results of the research of Earl Geddes about masking & IMD distortion & distortion audibility:
 
Top Bottom