• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Finally, you can calibrate your headphones to your ears.

For BCR we already have an EarEqual Reference Target, defined at the same blocked-ear reference plane where BCR measures, at the blocked ear-canal entrance. It is a first version, and I currently expect it to work best with open-back headphones.

The downstream ear-canal transfer is deliberately left outside both the BCR measurement and the target. When BCR is removed and you listen normally, your own ear canal is physically back in the chain, so those resonances return naturally rather than being imposed as an average target feature.

So the difficult HF question is not really how far up the target can remain valid before the ear canal starts to matter. The downstream canal acoustics are already excluded by the choice of reference plane.

The harder region is the upper treble, where pinna and concha variation, exact microphone position, headphone position and measurement repeatability all become increasingly important.

IEMs are different. An IEM sits inside the ear canal, so one should not transfer a blocked-ear headphone target directly to an IEM measurement.

If someone were to measure their own BCR (DF)HRTF surely these measurements would include concha and pinna variations that might be relevant to make IEMs sound more natural as well.

Most IEMs are tuned to something like the BK 5128’s DFHRTF for example. If we have the BK 5128’s BCR DFHRTF then can we extrapolate our own pinna and concha variations from the “average”.

So this is kind of my question.
 
Insertion depth and seal of an IEM matters.
So does the geometry of the ear-canal (diameter, length, shape etc.) differs from person to person which is not taken into account when measuring the entrance of a blocked ear canal.
 
Last edited:
I'll try to order these mics in the next month or so.
Could be interesting, but have to read some information about blocked ear canal vs open ear canal measurements.
 
The downstream ear-canal transfer is deliberately left outside both the BCR measurement and the target. When BCR is removed and you listen normally, your own ear canal is physically back in the chain, so those resonances return naturally rather than being imposed as an average target feature.
Yes, the ear canal will be “in the chain“ when llistening, but of course the measurement/recording situation at the ear entrance is different from the listening situation.
During measurement/recording there will be reflections from the blocked area that are different from the “reflections“ that occur during listening. This will create deviations between these situations, in particular in HF.

Nevertheless, the results from Møller et al (https://vbn.aau.dk/ws/portalfiles/portal/227875204/1995_M_ller_et_al_AES_Journal_b.pdf , see i.e fig 15) give a lot of confidence in blocked ear canal results for individual EQ.
And this is my experience, too.

The harder region is the upper treble, where pinna and concha variation, exact microphone position, headphone position and measurement repeatability all become increasingly important.

So far I measured Moondrop PARA 2 and HD480pro in my ears (all with the same PUI 6mm capsule)
Left is the result of three measurement runs (taking HPs off and on again) for both ears and both headphones.
Consistency is quite up there, differences between left and right ear look greater than measurement deviations.
[I did not remove and reinsert the microphone though. But these were quick kitchen table measurements without careful procedure.]
On the right the results of HD480pro are again in comparison with a second measurement (right ear only) some days later (the FR correction of all curves is different than in the left pic.)

Kopfhörer.png
HD480_R_two measurements..png


Interestingly, after EQ to the same target, the open headphone (PARA2) somehow sounds a bit more convincing than the closed HD480pro. However, this is not an easy comparison, as the closed HP “feels“ different in many ways.
 
So - I stumbled across this thread EXACTLY at the right time. And I've just ordered the mic's and the recommended adapter.
Initially - I have a Health&Safety test in mind - NOISE analysis in Motorcycle helmets.
I already suffer from Tinnitus for a whole gambit of reasons and yes! - I do wear obturating earplugs now.

Would anybody be able to recommend a simple recording app for use on android?
 
Yes, the ear canal will be “in the chain“ when llistening, but of course the measurement/recording situation at the ear entrance is different from the listening situation.
During measurement/recording there will be reflections from the blocked area that are different from the “reflections“ that occur during listening. This will create deviations between these situations, in particular in HF.
Yes, the blocked and open conditions do produce different reflections at the ear-canal entrance, or in Møller’s terms, different pressure division. But that does not automatically create an EQ error.

In the blocked case, both the headphone response and the target are defined with the same blocked boundary condition. In normal listening, both are subject to the open-ear condition instead. For an FEC headphone, the change from blocked to open affects both in the same way, so the ratio between target and headphone response, and therefore the required EQ, is preserved.

This is essentially the PDR/FEC issue in Møller’s formulation. The downstream transfer from the canal entrance to the eardrum is a separate part of the system and comes back naturally when the ear canal is no longer blocked.
 
I recently bought the Virtual Monitoring Pro extension to Sonarworks Reference. This includes in ear mics which are used to measure both a pair of speakers and a headphone. Their backend then creates a transfer function to simulate the speakers, and this is applied by the local software.

The process is quick and painless and the results work well overall. Without head tracking I don't get persistent frontal localisation, but sounds are placed in a 3D field. The main drawbacks are the necessity to use a Windows or Mac PC to run the software and the large decrease in volume.

I'll play around a bit with the in ear mics to see what I can do manually, though a first try yielded somewhat cryptic results. I may yet end up ordering EarEqual mics to compare and see if their measurements make things easier.
 
Sorry to budge in with partial off-topic (albeit it concerns the topic IMO) and this question actually has haunted me for a frew days now: how is it possible to reconcile personalized BRIRs with "taste"? I can understand that one might have a preferred "timbre tilt" to the sound, yet it feels quite inconsistent on theoretical side. Pardon me if this is explained elsewhere and published.

If one makes a personalized response (e.g. based on in-ear mics to match in-room speaker HRTF), then how does ear discern timbre coloration from HRTF mismatch? I mean, when one hears off-timbre speaker in the room it will always sound like off-timbre speakers, yet mismatched HRTF causes various types of mislocalized sound and other shenanigans with headphones, not "only" off-coloration. Is it simply the matter of HRTF shape matching the person, as in dips & peaks of whole FR, or what is the magic here?

I hope I got the point across, but basically TL;DR is - in pure theory, can one achieve perfect personalized FR on headphones which realizes off-timbre speakers in a room while not causing any spatial issues? In all honesty, if I got an "you're mixing things here to a big extent" answer then that also is reasonable.
Sorry for the delayed reply. I actually wrote a response earlier and then managed to mess up the posting.

Sean may have a better answer regarding preference research, but in pure linear theory I don't think there is a contradiction.

A personalized BRIR describes how a source at a particular position, in a particular room, reaches the listener's two ears. If the loudspeaker or source has a tonal coloration, that coloration is simply part of the signal passing through the BRIR.

That is also what happens with a real coloured loudspeaker in a room. If the headphone system reproduces the same two ear signals accurately, the loudspeaker can sound coloured while still retaining the correct spatial cues.

The ear does not necessarily have to separate the signal explicitly into "source timbre" and "HRTF". It interprets the complete combination of interaural cues, pinna related spectral cues, room information, head movement and learned expectations. Real sound sources have very different spectra, yet we can usually still localize them.

One important qualification is that a personalized magnitude frequency response alone is not enough. Spatial reproduction requires the personalized binaural transfer, including timing, phase, level differences and monaural pinna cues. A preferred tonal tilt can then be applied to the virtual source before the BRIR rendering.

Arbitrary separate left and right EQ, or narrow corrections applied without regard to the binaural transfer, could alter spatial cues.

So in pure theory, yes: an accurately personalized binaural system can reproduce an off-timbre loudspeaker without introducing spatial errors. If the BRIR was measured from that actual loudspeaker in that actual room, its coloration is already part of the measurement.
 
Hello everyone,

the EarEqual has arrived, sadly i have no good headphone here to test it out right now, sold everything before the new stuff arived, got only the Maxwell 2 to test it out, which will be sold down the line as well.

Setup

Damn, as someone without any experience in measurements and such tools/apps, it was a disaster to start, i needed a couple of hours, before i was sure the measurements were somewhere in the right ballpark. The correct loudness, so backgroundnoise is no problem, the need to choose left/right at two instead of just 1 position, how deep to insert the mic's, how to see the graph at all and not just part of it, how to match target and measured and how to combine it and so on Haha

I also couldnt just use the .csv from EarEual, it just didn't work, so i used AI to give me the 5 provided presets and the AI screwed it as well and fixed that in the second run, i probably could have just downloaded it somewhere else i guess.

Well, long story short. The Maxwell is not my cup of tea, not bad, but it just dosnt sound very mature and it won't wow me, even the EQ won't fix this, i just guess the quality of the loudspeaker is not so good, there has to be difference i dont know of. But the EQ made it sound much better, thats for sure.

This is the avg combined left/right with EarEqual as Target Reference

Maxwell 2 Target_AVG_BeforeCorrection.jpg



It may dont look like much, but it is indeed very different in sound, much more than i expected. This is the measurement after EQ with soundsource, i had to use bandwith instead of Q values. (AuneEQ Mac OS)

Maxwell 2 Target_Target_Real.jpg


- More Bass, sounds cleaner
- Mids stand out more a little more
- Treble sounds much more refined, it reminds me more of a high end Headphone without reaching it's quality. But the overall tonality is much better

But, i doont really know if its too much fatigue to have this kind of detail all the time, its really engaging but i could probably not listen too long with this. The Harman seems to be even more agressive and with oratory's profiles i always had to town down treble and bass, so i guess even the EarEqual Target is a bit to agressive for me personally, even thou it sounds really spectacular in terms of details, but you can also find more flaws in audio, so its a double sided coin.

I somehow guess thats why Hifiman tends to have this 2K dip? I need to find out, how i can create my own artificial target base on EarEquals Target with a bit less bass/treble and probably that same 2K dip, maybe thats the way to go for me.

What i can say for sure though is, that i am quite happy with the purchase and experience i get and the possibility to tune every headphone to my liking. Now i am much more happy with my purchase decision of the Meze Empyrean II as well, because looks and fit is so much better compared to the suvara. The Susvara may be near optimal for me already, but if i can get that or even better with the meze, its a win win for me.

Thank you very much for all the hard ork you put into this, Also looking forward for the EarEqual App that potentionally makes everything easier. Ive got a much better idea, which i had for years already!

DAC/AMP for example from Qudelix with 20 Bands PEA (5K got that already), but in this case for each ear 20 bands for separate EQ with provided Mic's from you and built in APP that does the measurements + EQ with different targets automatically with adjustments you can choose yourself such as Bass, Mids, Treble. PLUS hearing test, to include this into the EQ as well. That would be one hell of a DAC/AMP or DAP, i would buy something like that for 2-3k grand even without thinking.
 
I just received them and trying to get some measurements going, but I realised that you don't really offer a calibration for the delivered mics?

Is this something we're supposed to do ourselves?

Edit: the left and right microphone measure remarkably identical (like 1db tolerance) without any calibration curves.

Edit 2: Got the calibration files from Marko, the email was in my junk folder
 
Last edited:
After an hour or so of playing with the microphone, I have completely lost faith with subbass measurements of a GRAS fixture.

Most headphones seem to have a relatively terrible seal on my head.

I now understand why the 5128 “undermeasures” subbass. It’s because headphones seal to GRAS fixtures as if it has hot glue
 
Okay one more thing to mention that maybe is not super visible here in this thread or on the website.

This thing does not block/seal your ear canal

This imo makes a ton of sense (specially now re-reading the thread) and the measurements repeatability is incredibly high because of it.
 
Last edited:
Was also surprised that it didnt blocked entirely and i have no clue about this topic, was reading a lot already, but understanding things is another story.

I measured on 2 heads and the most noticeable difference were in the bass below 300Hz and treble above 1kHz. The headphone we used was not sealing the second person head because its simply too big and because of glasses. Also the measured outcome after eq was relative close to the predicted eq. I did all the measurements with -12-20) dbfs, which was quite uncomfortable. Not sure if i should start the measurements from 20 or 0Hz as starting point. Otherwise the measurements almost looked identical all the time.

Next time i get the headphones deliverd that i actually want listen to, iam gonna do 10 measurements with slight different position of the headset and combine them before starting to eq.

Any recommendation for autoeq settings within REW or on autoeq website? Depending on the settings on the autoeq website it sounds quite different.
 
Thanks everyone. There are several really useful observations here, so I’ll try to answer them together.

First, @GalaxyRadio , your Maxwell result is very encouraging. Nice that you preferred the sound after EQ, and that your remeasurement followed the EarEqual target quite closely.

Your difficulty with the setup is very useful feedback, thanks. The current REW workflow is a bit complicated for someone new to measurements. That is exactly the problem the EarEqual App is intended to solve.

One correction from our side: the target download on the website previously contained several target/reference curves in the same CSV. That was our mistake, and it has now been fixed. The download now contains only the EarEqual Reference Target v0.1.

On microphone calibration, each finished BCR is indeed individually calibrated and sent by email with separate left and right calibration files. Sometimes emails end up in junk.

Regarding seal and bass, I would not lose faith in GRAS or 5128 measurements. Standardized fixtures are useful because they give us a controlled comparison. But what you are seeing is also why measuring on the actual listener might be better. The real headphone seal on a real head can differ a lot from the fixture seal because of glasses, head shape, pad condition, clamp force and so on.

A clarification about BCR itself: it is not a perfectly rigid, and not always a perfectly sealed ear-canal block. How much it seals or leaks depends on the individual ear geometry and how it sits in the ear.

If the leakage is mainly a low-frequency phenomenon, the dimensions of the BCR, leak path and ear entrance are extremely small compared with the acoustic wavelength. The local pressure field is therefore approximately uniform, so a low-frequency leak does not necessarily produce a significant spatial pressure measurement error.

If the target is measured with the same BCR, the effect is common to both measurements and cancels in Target / Headphone Response. With a universal numerical target such as EarEqual Reference Target, the exact measurement condition matters more. I would be much more concerned if substantial leakage or resonant behaviour extended into the mid or high frequencies than if it were mainly a low-frequency effect.

For the sweep settings, start at 20 Hz. There is no reason to include DC in a headphone calibration sweep.

The measurement also should not be uncomfortably loud. dBFS is only the digital signal level and does not tell us the SPL at the ear by itself. The aim is simply to use enough level for a good signal-to-noise ratio. With the microphones sitting very close to the headphone, there is normally no reason to make the sweep unpleasant.

On averaging measurements, repeated reseats is a good idea, but I would not deliberately create ten very different headphone positions and average everything blindly. I would rather make several realistic reseats representing how you normally wear the headphone, inspect the spread, and average measurements that belong to the normal wearing condition. Otherwise it is easy to average away real positional behaviour.

For AutoEQ settings, I do not yet want to recommend one universal preset. The main rule I would use for now is to correct broad, repeatable deviations, but not aggressively fill narrow, deep notches, especially in the upper treble. We are working on regularization in the App specifically around this issue. Different AutoEQ settings can indeed sound very different because they make different assumptions about smoothing, boost limits and notch inversion.

And @AMDfanTO, looking forward to seeing your measurements when you get some time with the BCR.’

Thanks!
 
Picked up the mics and did a quick profile on my Hifiman Edition XS. I'm extremely satisfied with the results. The sub bass and ear gain measurements were surprising.

I am using after market ear pads that have an attached foam insert that definitely warm things up, and potentially reduce the sub bass region.

Using your provided target yielded a smooth, dynamic and very speaker-like sound. Thank you so much for bringing this to market.
 

Attachments

Last edited:
Ok, just ordered microphones :)
Hope to receive them soon, gonna be fun.

ps. I'm using topping ex2 interface and want to use earequal with it.
I have rode vxlr+ (need to buy second one) adapter so will switch 48v to ~4V, but I'm sure I also need spliter so I can measure both channels right?

Something like this:
1793176821749.png


3,5mm female stereo to 2x 3,5mm ts mono
 
Last edited:
Frequency Response Graph.png



This picture is basically how the Hifiman HE400SE measures on my head with the microphone provided here vs Amir's GRAS/KB5000 fixture.

To be honest I'm very shocked at how big the difference is, I already experimented with some EQ to see the impact of filling up the presence region. the subbass was no brainer to fill up and sounds great when I did it. I would argue that the microphones are worth it for the sub bass improvements on their own.

For people who had unequal ears from a very young age (so not induced by neurological damage), I think these measurements might also give some input on how to fix this. but this is very anecdotal at the moment, I need some time windowed measurements of my speakers to see how my ears measure.
 
I think these measurements might also give some input on how to fix this
The question is if that would need fixing (your brain is used to getting its input that way).
And if that can be fixed at all, especially when the issues are most likely not for acoustic reasons but medical.
 
The question is if that would need fixing (your brain is used to getting its input that way).
And if that can be fixed at all, especially when the issues are most likely not for acoustic reasons but medical.
Very good point.
 
Back
Top Bottom