AKG N700nc m2 Measurements

[MatthewS]

Just one point, the dip in Amir's graph that you're pointing to is at 6kHz, whereas your EQ peak is at 8kHz, so you're not filling that gap.

Umm. Center frequency is 5.5k. That is what I wrote.

My original post was incorrect. I couldn’t remember.

 

[Robbo99999]

I got them for $299 new on clearance a few years ago

As I added to my previous post, Innerfidelity's Tyll measured TruNote and all the spatial modes and bass/treble tone controls (not on a GRAS though, on the same HATS Rtings use), see here:

AKG N90Q Noise Canceling Auto-Calibrating Over-Ear Headphones Page 2

Electronics The AKG N90Q is a fully active electronic headphone; it can not be used passively when the battery runs out after about 10 hours. That included storage battery will come in handy on trans-oceanic flights.

www.stereophile.com

Ah, so TruNote sorted out the channel imbalance! And perhaps made them slightly less peaky overall, although small effect, but the channel matching effect was very noticeable. (Good channel matching is worth it alone, in my experience through measuring my headphones on miniDSP EARS and then creating a perfect channel match through the whole frequency range on all my headphones.).

 

[moosso]

Actually many other well received headphones have a dip there e.g. HifiMan models, and...the Sennheiser HE-1 This dip pushes back instruments/voices with these frequencies in the mix and can create a sense of more spaciousness (particularly depth), as people like @jhaider found with the N700 M2.

As this video said some mixing and mastering engineers use nearfields speakers that will cause few dB cancellation around 2KHz, if they compensate it these track will have more 2KHz energe than it should have.
Headphones with a 2KHz dip could sound better with those tracks.

 

[LightninBoy]

Aren't these the headphones Sean Olive says he uses for casual listening?

I bought a pair of these a few years back when they were being coupled with a new samsung smart phone and everyone hated them because of connection issues. I think I paid $70.

For some reason, I seem to always buy HPs with bad channel match. (I also bought the NAD HP50). I never trusted these HPs because they make everything sound rich and lush. No edge at all. So the measured FR response aligns to what I'm hearing.

But the sound is not the problem with these. Its the user experience. The controls are tiny and difficult to use. The mobile app is a joke. The HPs can connect to multiple sources but when it does its constantly glitching (I suspect this was why folks hated them). It has a lot of features but nowhere are they fully explained. For example, there's an EQ built into the app - great. Is that available always, only in BT mode? Plugged in? Only plugged into the phone? Only when the app is active? There are two noise cancelation overide modes. Nowhere are they fully explained. And on top of all that, the HPs feel claustrophobic on your head with tiny cups pressed very tight.

They are fine at the blowout price I got them. But otherwise, I do not recommend.

 

[GaryH]

Aren't these the headphones Sean Olive says he uses for casual listening?

I believe he personally owns the original, not this M2 version.

 

[jhaider]

But the sound is not the problem with these. Its the user experience. The controls are tiny and difficult to use. The mobile app is a joke. The HPs can connect to multiple sources but when it does its constantly glitching (I suspect this was why folks hated them).

Do people actually use the local controls on headphones? I’ve found they all suck - AirPods Pro included. I couldn’t even tell you what the controls on these headphones are.

As for multiple (2) connections it seems to work well enough for me. That’s why I bought two pairs of these - one pairs with my MacBook and iPhone, the other with my work computer and iPhone. Nobody else has anything remotely as good as Apple’s Continuity, but unfortunately organizations don’t all use Macs so even those of us with better taste personally are often relegated to using clunkers.

 

[lewdish]

Its just a cheapo harman tuned BT headphone, not much exciting, works fine for its target demographic, but sad to see this is what AKG is today, and what it will look like maybe a decade from now. Im hoping the new team at Austrian Audio will inspire us like AKG used to~ Digging the their x65 so far

 

[GaryH]

Cheap is a good thing. Here's another good cheap offering from AKG:

https://twitter.com/x/status/1578469880696696832

The more than twice as expensive Austrian Audio X65 however:

 

[usersky]

Its just a cheapo harman tuned BT headphone, not much exciting, works fine for its target demographic, but sad to see this is what AKG is today, and what it will look like maybe a decade from now. Im hoping the new team at Austrian Audio will inspire us like AKG used to~ Digging the their x65 so far

The Austrian Audio equivalent of N700m2 is Hi-X25BT. Another level entirely IMO, it melts on your ears and sounds very good with no EQ. I'd love to see it measured. I also have Hi-X65 but it's little sister beat the flagship comfortwise.

 

[solderdude]

Austrian Audio HI-X15 & HI-X25 BT, 밀폐형 헤드폰 측정 리뷰 - 영디비

Austrian Audio에서 10만 원대 엔트리급 밀폐형 헤드폰을 출시했습니다. HI-X25 BT 블루투스 탑재 헤드폰과 HI-X15 유선 헤드폰인데요! 이번 리뷰는 Impedance 그래프에 대한 내용이 꽉꽉 담겨있습니다. Specificatio...

www.0db.co.kr

Not very Harmanesque.

 

[solderdude]

K702 acc.. to Oratory

different fixture:

Headphones v1.5 - Graph Tool

www.rtings.com

Crin CA43AG-7:

AKG K702 – In-Ear Fidelity

crinacle.com

 

[Koeitje]

I'm curious as to what percent of Bluetooth headphone purchasers would ever use them in wired-mode? I saw this and said oh good, a wireless headphone measurement. Guess again...

This question is almost worth a poll, since the propensity here seems to be to measure BT phones in wired-mode. (Although, I do understand this is because the measurement gear is designed for wired designs, primarily). My completely unsubstantiated guess is that about 1% of BT headphoners use them hard-wired. I see the time spent on wired measurements of wireless headphones as somewhat pointless.

How well do BT and wired performance equate, in general? I'll admit, I don't follow most headphone disussion here, because most is about wired 'phones.

Remember, ASR tries to be helpful to others. This quick, problematic, inconsistent with others, measurement, really didn't need to be a front-page discussion point. It may shy buyers away from what may be a perfectly fine, not expensive, BT headphone. Despite our Harman headphone bashers, AKG is a renown headphone brand. Half the daily traffic here is new visitors that look for information about specific devices. Hopefully, they read that is was measured wired, which is not a natural assumption to be made when looking for info on wireless headphones.

I only use the wired mode when I plug it into an Xbox controller with an inline mic in between. Its probably 0,01% of my total usage that I use it wired.

 

[Blorg]

I’ve seen statements like this here as well as on other Bluetooth headsets like the Bose nc700. I have a hard time understanding how this works. Do the headphones use an ADC to put the voltage back in the digital domain, then apply DSP and output the result with its built-in DAC? Or is DSP being used as shorthand for active amplified alteration of signal?

Basically, yes. It's redigitized and fed through the DSP and then DAC and amp in the headphone. This is also why they distort so much at 114dB, it's overloading the ADC. And why others like the Mark Levinson have such high measured impedance, it's basically a line in.

This what I do with them, the EQ app makes it look more than it is but it's basically something similar to -2dB low shelf from ~350Hz. Subjectively, this is enough for me, any more and they are getting a bit thin. They sound very good to for BT headphones, excellent for the $40 I paid for them.

 

[MayaTlab]

Hard to tell how Crinacle's / Resolve's / Amir's all line up without putting them all on the same graph.

Normalised at 500Hz for the first graph, and across 3 octaves centred at 500Hz for the second :

Error curve, normalised at 500Hz, 50-10k Hz :

Inter-individual variation on a cohort of test subjects in wired, passive mode, with blocked ear canal entrance microphones (not comparable to ear simulators measurements !) :

https://danishsoundcluster.dk/wp-content/uploads/2022/11/Olive_DSD_2022.pdf

You can discard the lonely trace most different from the others, it's an error on Harman's part.

Because the traces were normalised based on the average level between 200-1k Hz, it isn't possible to know what it would look like when engaging the feedback mechanism (DSP, active mode) above the range where the latter operates, as while that system will try to reduce variation at low frequencies (for a feedback ANC system, a specific digital value means a specific SPL output up to x frequency, within the system's capabilities to correct for deviations), we don't know what the actual SPL is across the spectrum for each individual.

That trough in the 1.5-3kHz range is really going to negatively impact the listening experience for sure.

There's a pretty good chance that I experienced in situ a higher response around 2-3kHz than what the graphs would suggest.

I have for a while tried to guesstimate what the Harman target is meant to sound or measure like on my own head, with in situ microphones of various kind, and if I average a number of large, open over-ears (HD650, HD58X x2, HD560S, Sundara, HE400SE, Clear MG), measured with blocked ear canal entrance microphones, after equalisation to the Harman target using Oratory's measurements, I get the dark blue line here :

As usual with blocked ear canal entrance measurements, don't compare these to any other, simply observe the relative differences between the traces.

The dotted black line corresponds to a pair of AirPods Max more strictly equalised to Harman at low frequencies, where its feedback system and very low sample variation ensures that it rather faithfully delivers the frequency response that's been measured on ear simulators, as a confirmation that the average corresponds quite well to what the Harman target is meant to measure like.

Another point of comparison could be to measure in situ a bunch of "Harman target tuned" passive HPs from Harman, here two samples of the Mark Levinson 5909, K371 and 710BT. All of these are susceptible to leakage, so at lower frequencies they'll struggle to have a constant response across individuals and ear simulators, the overall average remains too heavily influenced by the individual features of each model, and they don't strictly adhere to the target to begin with. Take it less as an illustration of what the Harman target is meant to measure like in situ with this method, but rather as a point of comparison with how a bunch of headphones Harman considers to sound good out of the box measured on my own head.

I also believe that I prefer a slightly more relaxed response in the 1-4kHz than the average obtained using Oratory's measurements, probably a bit closer to the average obtained using Harman's headphones, but this is just my unsubstantiated subjective impressions so don't take them too seriously.

Using these averages, here's how my sample of the N700NCM2 measured in situ in comparison :

Now this is quite the trace arithmetic acrobatics and probably a step too far, but if I plot the "error" curve of my ex-N700NCM2, in situ, compared to these two averages (because the average for Harman's headphones is evidently wrong at low frequencies, I've merged it with the AirPods Max curve at around 500Hz), and compare it to the error curves derived from ear simulator measurements, vs. the Harman target, we get this :

Since we're on ASR I've highlighted the site's measurements.

Why the difference ? I don't know. The N700NC M1 and M2 are quite susceptible to significant variation in SPL in the 1-3kHz region when testing them under pad compression, but I can't say that this sort of test always neatly aligns with the discrepancies I observe between ear simulators and how a pair behaves on my own head. It could also be related to geometry around the pinna, most ear simulators are inaccurate in that regard and the N700NC M1/M2 sit at a very different angle relative to the pinna, on my head, than on ear simulators with a flat plate around the pinna.

given we know this isn't proper parametric EQ being implemented, but instead Graphic EQ which has a bandwidth of Q1.41 for each of the 10 graphic EQ bands, then this is not quite so bad as they are broader filters - but either way the EQ is not doing what you said it does, it's not filling in the dip in Amir's graph.

So I think that it's a bit more complicated than that.

For a start it varies depending on the headphones model. I have at home the JBL Live Pro 2 and N700NC M1, and I've had various other Harman headphones over time, including the N700NC M2, and if I try (emphasis on "try" as the app's UI makes it difficult to nail it) to input the same values I often get different results.

More importantly I have yet to fully understand how the values you input translate into a measured result. It is not a straightforward graphic EQ. You get a sense that it does try to map the UI's curve to 10 fixed bands, but the underlying logic remains quite mysterious to me.

A few examples using the N700NC M1. I recall that the UI is similar on the N700NC M2, but I don't know whether the results would be (all I know in regards to the N700NC M2 is that I wasn't able to solve the 800-4000Hz range to satisfaction with it).

First profile (dB level annotation mine, it's approximate as the UI only tells you the dB level while you drag the point and only in 1dB increments) :

Without smoothing you see the individual bands showing through, but even with a lot of smoothing it seems to overshoot a bit the 6dB difference I tried to enter.

Second :

This time the 6dB difference seems to have been hit, but the tail at both ends of the curve extends a bit further than what the app's UI suggests.

Third :

Applying a negative EQ in the band next to the 1kHz one dragged down the difference in magnitude at 1kHz, despite the UI showing no change. This suggests to me a limited maximum slope, but again it's not clear exactly what that is.

Fourth, purposefully aligning the peak in between two bands :

For this one it clearly missed the 6dB difference.

My take on Harman's app EQ : the chance that the EQ profile you've entered is anywhere near the actual FR difference between no EQ and EQ profile applied is really low and it's only really useful if you have some measurement device to actually know the resulting difference (a cheap flat plate will do for that task). Otherwise it's just misleading. Without measurements available, maybe only stick to very low Q tilts or low magnitude adjustments.

Amir tested the N700NC M2 wired and "on". As he indicated, any DSP correction would be active. In the case of the N700NC M2, that would mean ANC and, if enabled, ambient/talk-thru mode, but no EQ. This could be a problem for the measurements... For example, the ANC processing has a time-element factor (like a PID-type response). This could very well explain the difficulty Amir had to match L & R channels and the choppy FR response...

Would you mind clarifying what you mean by that ?

Some headphones need priming before being measured with a sweep (Airpods in some situations among others), but classic ANC systems probably not. I have measured the N700NC with both sweeps and noise and found no difference.

The original also had the Trunote feature which I think they dropped for the M2 - personalised frequency response calibration for your specific ears via playing back and recording a sine sweep while wearing them. Would be very interesting to see how that measures with/without Trunote engaged on a test rig.

Wow, they're some expensive not quite old skool headphones! (launched 2015). TruNote would be pretty darn interesting to see measured on GRAS - stock vs TruNote on GRAS. GaryH mentioned he'd like to see such measured, would be interesting to see if anything could be worked out from measuring the effect. Maybe if it was measured on different rigs (GRAS / B&K / miniDSP EARS, etc), I wonder if any patterns (& insight) could be deduced from measuring stock vs TruNote.

So, TruNote . This one is opening a whole can of worms.

I feel that TruNote is misunderstood (it's not really about individual personalisation), and to be frank I will readily admit that I also don't really understand what it's trying to do !

It's based on an article from Ulrich Horbach, "Characterizing the Frequency Response of Headphones – a new Paradigm", which is behind a paywall but isn't really worth your money (PM me if you want to talk about it).

Mad_Economist would be quite a bit more eloquent than me on what's problematic with that article. Be warned, it includes a lot of stuff you aren't used to see from Harman's articles, including DIY flatplates :

The way I understand it is that it considers the plane above the driver a reference plane that should be equalised to a common target across individuals, regardless of fit / coupling, and then leaves everything else to fall where it may beyond that plane.

To do so, all TruNote headphones are featuring at least two electrets slightly spaced apart above the driver (instead of the usual one you see with ANC feedback systems), ex here a N90Q prototype featured in the article, the actual N90Q, and the N700NC M1 that I own :

When performing the TruNote calibration, a sweep is played and then an EQ profile is applied to a common target at that plane.

So, what does it do with my own N700NC M1 ?

First of all, you can't reset it without doing a full factory reset on these, by pressing these buttons for a few seconds :

Using the same averages as reference as I used before, here's how my pair measures before and after TruNote :

It seems to have corrected for some issues, but I don't really know on which grounds exactly and whether the improvements are intentional or not. Ideally you'd test it on a cohort of individuals with in situ measurements, but that wasn't done in the article mentioned above.

It also did introduce a slight channel imbalance, when performed in situ, on my own head, below 1kHz, for some strange reason I can't really explain - even more so given that it is in the range where the feedback system operates (the above is the average of the R and L channels).

I haven't found a lot of Harman compliant measurements of the N700NC M1 other than this one from Sean Olive / Harman, and I don't know if it was performed before or after TruNote calibration :

My gut feeling is that headphones such as the N90Q or N700NC M1 may be more prone to coupling issues, and that in some situations TruNote may help to compensate for it and align the in situ response closer to an average baseline above the driver's plane, but I also feel that it might introduce errors and in no way is it capable of truly individualising the response.

 

[nathan]

Picked up a pair of the first version for $99. I haven’t heard any ANC Bluetooth over the ear cans near that price point that follow the Harman curve better.

(About the ANC. It is adequate to remove normal household background noise but on an airplane, subway or around power tools I’d prefer Sony, Bose, or Apple.)

 

[Robbo99999]

Normalised at 500Hz for the first graph, and across 3 octaves centred at 500Hz for the second :
View attachment 292129View attachment 292130

Error curve, normalised at 500Hz, 50-10k Hz :
View attachment 292134

Inter-individual variation on a cohort of test subjects in wired, passive mode, with blocked ear canal entrance microphones (not comparable to ear simulators measurements !) :
View attachment 292131

https://danishsoundcluster.dk/wp-content/uploads/2022/11/Olive_DSD_2022.pdf

You can discard the lonely trace most different from the others, it's an error on Harman's part.

Because the traces were normalised based on the average level between 200-1k Hz, it isn't possible to know what it would look like when engaging the feedback mechanism (DSP, active mode) above the range where the latter operates, as while that system will try to reduce variation at low frequencies (for a feedback ANC system, a specific digital value means a specific SPL output up to x frequency, within the system's capabilities to correct for deviations), we don't know what the actual SPL is across the spectrum for each individual.



There's a pretty good chance that I experienced in situ a higher response around 2-3kHz than what the graphs would suggest.

I have for a while tried to guesstimate what the Harman target is meant to sound or measure like on my own head, with in situ microphones of various kind, and if I average a number of large, open over-ears (HD650, HD58X x2, HD560S, Sundara, HE400SE, Clear MG), measured with blocked ear canal entrance microphones, after equalisation to the Harman target using Oratory's measurements, I get the dark blue line here :

View attachment 292136

As usual with blocked ear canal entrance measurements, don't compare these to any other, simply observe the relative differences between the traces.

The dotted black line corresponds to a pair of AirPods Max more strictly equalised to Harman at low frequencies, where its feedback system and very low sample variation ensures that it rather faithfully delivers the frequency response that's been measured on ear simulators, as a confirmation that the average corresponds quite well to what the Harman target is meant to measure like.

Another point of comparison could be to measure in situ a bunch of "Harman target tuned" passive HPs from Harman, here two samples of the Mark Levinson 5909, K371 and 710BT. All of these are susceptible to leakage, so at lower frequencies they'll struggle to have a constant response across individuals and ear simulators, the overall average remains too heavily influenced by the individual features of each model, and they don't strictly adhere to the target to begin with. Take it less as an illustration of what the Harman target is meant to measure like in situ with this method, but rather as a point of comparison with how a bunch of headphones Harman considers to sound good out of the box measured on my own head.

View attachment 292137

I also believe that I prefer a slightly more relaxed response in the 1-4kHz than the average obtained using Oratory's measurements, probably a bit closer to the average obtained using Harman's headphones, but this is just my unsubstantiated subjective impressions so don't take them too seriously.

Using these averages, here's how my sample of the N700NCM2 measured in situ in comparison :

View attachment 292138

Now this is quite the trace arithmetic acrobatics and probably a step too far, but if I plot the "error" curve of my ex-N700NCM2, in situ, compared to these two averages (because the average for Harman's headphones is evidently wrong at low frequencies, I've merged it with the AirPods Max curve at around 500Hz), and compare it to the error curves derived from ear simulator measurements, vs. the Harman target, we get this :

View attachment 292139

Since we're on ASR I've highlighted the site's measurements.

Why the difference ? I don't know. The N700NC M1 and M2 are quite susceptible to significant variation in SPL in the 1-3kHz region when testing them under pad compression, but I can't say that this sort of test always neatly aligns with the discrepancies I observe between ear simulators and how a pair behaves on my own head. It could also be related to geometry around the pinna, most ear simulators are inaccurate in that regard and the N700NC M1/M2 sit at a very different angle relative to the pinna, on my head, than on ear simulators with a flat plate around the pinna.



So I think that it's a bit more complicated than that.

For a start it varies depending on the headphones model. I have at home the JBL Live Pro 2 and N700NC M1, and I've had various other Harman headphones over time, including the N700NC M2, and if I try (emphasis on "try" as the app's UI makes it difficult to nail it) to input the same values I often get different results.

More importantly I have yet to fully understand how the values you input translate into a measured result. It is not a straightforward graphic EQ. You get a sense that it does try to map the UI's curve to 10 fixed bands, but the underlying logic remains quite mysterious to me.

A few examples using the N700NC M1. I recall that the UI is similar on the N700NC M2, but I don't know whether the results would be (all I know in regards to the N700NC M2 is that I wasn't able to solve the 800-4000Hz range to satisfaction with it).

First profile (dB level annotation mine, it's approximate as the UI only tells you the dB level while you drag the point and only in 1dB increments) :
View attachment 292127View attachment 292126
Without smoothing you see the individual bands showing through, but even with a lot of smoothing it seems to overshoot a bit the 6dB difference I tried to enter.

Second :
View attachment 292121View attachment 292125
This time the 6dB difference seems to have been hit, but the tail at both ends of the curve extends a bit further than what the app's UI suggests.

Third :
View attachment 292120View attachment 292124
Applying a negative EQ in the band next to the 1kHz one dragged down the difference in magnitude at 1kHz, despite the UI showing no change. This suggests to me a limited maximum slope, but again it's not clear exactly what that is.

Fourth, purposefully aligning the peak in between two bands :
View attachment 292119View attachment 292123
For this one it clearly missed the 6dB difference.

My take on Harman's app EQ : the chance that the EQ profile you've entered is anywhere near the actual FR difference between no EQ and EQ profile applied is really low and it's only really useful if you have some measurement device to actually know the resulting difference (a cheap flat plate will do for that task). Otherwise it's just misleading. Without measurements available, maybe only stick to very low Q tilts or low magnitude adjustments.



Would you mind clarifying what you mean by that ?

Some headphones need priming before being measured with a sweep (Airpods in some situations among others), but classic ANC systems probably not. I have measured the N700NC with both sweeps and noise and found no difference.




So, TruNote . This one is opening a whole can of worms.

I feel that TruNote is misunderstood (it's not really about individual personalisation), and to be frank I will readily admit that I also don't really understand what it's trying to do !

It's based on an article from Ulrich Horbach, "Characterizing the Frequency Response of Headphones – a new Paradigm", which is behind a paywall but isn't really worth your money (PM me if you want to talk about it).

Mad_Economist would be quite a bit more eloquent than me on what's problematic with that article. Be warned, it includes a lot of stuff you aren't used to see from Harman's articles, including DIY flatplates :
View attachment 292142

The way I understand it is that it considers the plane above the driver a reference plane that should be equalised to a common target across individuals, regardless of fit / coupling, and then leaves everything else to fall where it may beyond that plane.

To do so, all TruNote headphones are featuring at least two electrets slightly spaced apart above the driver (instead of the usual one you see with ANC feedback systems), ex here a N90Q prototype featured in the article, the actual N90Q, and the N700NC M1 that I own :
View attachment 292145View attachment 292144View attachment 292143

When performing the TruNote calibration, a sweep is played and then an EQ profile is applied to a common target at that plane.

So, what does it do with my own N700NC M1 ?

First of all, you can't reset it without doing a full factory reset on these, by pressing these buttons for a few seconds :
View attachment 292148

Using the same averages as reference as I used before, here's how my pair measures before and after TruNote :

View attachment 292150

It seems to have corrected for some issues, but I don't really know on which grounds exactly and whether the improvements are intentional or not. Ideally you'd test it on a cohort of individuals with in situ measurements, but that wasn't done in the article mentioned above.

It also did introduce a slight channel imbalance, when performed in situ, on my own head, below 1kHz, for some strange reason I can't really explain - even more so given that it is in the range where the feedback system operates (the above is the average of the R and L channels).

I haven't found a lot of Harman compliant measurements of the N700NC M1 other than this one from Sean Olive / Harman, and I don't know if it was performed before or after TruNote calibration :
View attachment 292153

My gut feeling is that headphones such as the N90Q or N700NC M1 may be more prone to coupling issues, and that in some situations TruNote may help to compensate for it and align the in situ response closer to an average baseline above the driver's plane, but I also feel that it might introduce errors and in no way is it capable of truly individualising the response.

Either way, there's a shedload of variance going on with this N700!

 

[MayaTlab]

Either way, there's a shedload of variance going on with this N700!

Indeed, but I wouldn't necessarily attribute it to sample variation (actually I'm pretty certain that it isn't to that extent). Different measurement practices and fixtures could also explain that variance when combined with a design that is more sensitive to coupling issues. Interestingly, even though the N700NCM2 features a feedback system, it still shows quite a bit of variance at lower frequencies.

Compared to some other HPs (Bose 700, HD560S), using the same scale :

The HD560S is a good example of a pair of headphones that has both low enough sample variation and low enough sensitivity to coupling issues that most measurements tend to align pretty well, even though the fixtures or methodologies might be a bit different. The Bose 700 features a rather robust feedback system up to 800Hz or so, but also a volume dependent EQ that may contribute to a moderate degree to the differences in that range (but I don't think that it can explain everything). In the 1-4kHz range, and possibly above, it's very sensitive to coupling issues on the other hand.

 

[Robbo99999]

Indeed, but I wouldn't necessarily attribute it to sample variation (actually I'm pretty certain that it isn't to that extent). Different measurement practices and fixtures could also explain that variance when combined with a design that is more sensitive to coupling issues. Interestingly, even though the N700NCM2 features a feedback system, it still shows quite a bit of variance at lower frequencies.

Compared to some other HPs (Bose 700, HD560S), using the same scale :

View attachment 292614View attachment 292615View attachment 292616

The HD560S is a good example of a pair of headphones that has both low enough sample variation and low enough sensitivity to coupling issues that most measurements tend to align pretty well, even though the fixtures or methodologies might be a bit different. The Bose 700 features a rather robust feedback system up to 800Hz or so, but also a volume dependent EQ that may contribute to a moderate degree to the differences in that range (but I don't think that it can explain everything). In the 1-4kHz range, and possibly above, it's very sensitive to coupling issues on the other hand.

Yep, and that also goes along with Amir finding the N700 difficult to measure - ie varying a lot with placement. At the end of the day it's still variation that you experience, whether it's due to coupling issues or real unit to unit variation - so this particular headphone is one I would stay away from.

 

[markanini]

So what's the most consistent closed-back? And maybe an alternative in the lower three digits.

 

[markanini]

Sean Olive psoted his measurement of the original N700NC here: https://www.audiosciencereview.com/...a-and-dan-clark-aeon2-noire.20329/post-670627