Harman curve without ear canal? HD650, DT150_100P, DT770 Pro 250 Ohm, Bose QC25 measured

[ADU]

The Harman curve has alot of smoothing in the treble. And neither the Rtings nor the Harman targets are that accurate or reliable above a certain range in the treble and upper mids imho, beyond some very broad strokes. So I wouldn't put much faith in either of them, especially in that range.

With a little experimentation, I suspect you'll be able to get as good, if not some better results from your own home spun target compensation curves. Just my 2c fwiw.

 

[ADU]

To get some smoother headphone targets from the speakers, you might consider taking several different in-concha measurements, maybe with your head in slightly different orientations to the speakers each time (e.g. titled/angled +/-5 degrees in each axis, XYZ). And then computing an average response curve for each ear from those different samples.

You could also try sampling the responses over several slightly different seating/listening positions, as Harman did for some of their in-room measurements.

 

[Thomas_A]

I could not finish during this weekend but below are some intermediate results. It is evident that real-world sealing against the head when you have a lot of hair etc is an issue. There are small and big ears, no hair to big curly hair present.

HD650, three individuals:

DT770 Pro, four individuals:

DT150/100P, four individuals:

Bose QC25, four individuals:

 

[MayaTlab]

Bose QC25, four individuals:

Eager to get more impressions and observations on your end, but I was in particular looking for that, and it didn't disappoint :

 

[Thomas_A]

Eager to get more impressions and observations on your end, but I was in particular looking for that, and it didn't disappoint :
View attachment 202083

Thanks,

not sure what to make of this yet. While the larger-cups headphones seemed rather consistent up to 4-5 kHz (other than sealing issues), the QC25 did not. Perhaps it is the smaller cups of the Bose?

 

[MayaTlab]

Thanks,

not sure what to make of this yet. While the larger-cups headphones seemed rather consistent up to 4-5 kHz (other than sealing issues), the QC25 did not. Perhaps it is the smaller cups of the Bose?

I'd love to have your ideas on the matter. But I invite you to re-measure your QC25 on your own head with varying amounts of pad compression.

Difficult to be conclusive without more extensive repeated testing, but my guess is that one of the variables that varies when compressing pads is also present as a variable across individuals. Perhaps volume of air in the front volume, which could vary between individuals because of varying geometry around the pinna, something ear simulators can't test particularly well for (and none of which are anatomically accurate around the pinna to start with).

Some headphones, and in particular some active headphones, and even more so in particular some Bose/Sony/etc. headphones, are extremely sensitive in some part of the spectrum above the range where their feedback mechanism operates and below a few kHz or so to pad compression (while their feedback mechanism makes them, on the other hand, extremely insensitive to pad compression or small leaks in the range where it operates).

Some ideas thrown in these posts :

HBK Headphone Measurement Talks from Head-Fi and Sean Olive

As I said, design criteria is not proof of accomplishment. Independent testing and research needs to be performed to ascertain that. Are you suggesting that there are issues with B&K's underpinning research? It seems to agree well with a randomly grabbed paper from an unrelated source, and, of...

audiosciencereview.com

Mark Levinson 5909, new ANC wireless cans that advertise use of the Harman curve

Forgot to give credit to @Sean Olive in his tweet here. I believe this is an actual production sample - it slightly differs how it measures from the production validation sample posted earlier. What @Sean Olive thinks about Brent Butterworth´s measurements at Soundstage Solo...

audiosciencereview.com

These headphones also tend to show the highest amount of variation between ear simulators in that "above feedback range" - "below a few kHz" range.
Ex : QC35 in Sean Olive's presentation for the HBK conference :

Rtings also tested the 5128 against their current ear simulator and SamV released some files in the wild.

I've been as a result very wary of over-interpreting some headphones' ear simulator results without additional information on their behaviour under several variables. For headphones such as the Bose 700, QC45 or Airpods Max, I've found ear simulator results to be a rather poor predictor of their actual on-head behaviour for me.

 

[Thomas_A]

I'd love to have your ideas on the matter. But I invite you to re-measure your QC25 on your own head with varying amounts of pad compression.

Difficult to be conclusive without more extensive repeated testing, but my guess is that one of the variables that varies when compressing pads is also present as a variable across individuals. Perhaps volume of air in the front volume, which could vary between individuals because of varying geometry around the pinna, something ear simulators can't test particularly well for (and none of which are anatomically accurate around the pinna to start with).

Some headphones, and in particular some active headphones, and even more so in particular some Bose/Sony/etc. headphones, are extremely sensitive in some part of the spectrum above the range where their feedback mechanism operates and below a few kHz or so to pad compression (while their feedback mechanism makes them, on the other hand, extremely insensitive to pad compression or small leaks in the range where it operates).

Some ideas thrown in these posts :

HBK Headphone Measurement Talks from Head-Fi and Sean Olive

As I said, design criteria is not proof of accomplishment. Independent testing and research needs to be performed to ascertain that. Are you suggesting that there are issues with B&K's underpinning research? It seems to agree well with a randomly grabbed paper from an unrelated source, and, of...

audiosciencereview.com

Mark Levinson 5909, new ANC wireless cans that advertise use of the Harman curve

Forgot to give credit to @Sean Olive in his tweet here. I believe this is an actual production sample - it slightly differs how it measures from the production validation sample posted earlier. What @Sean Olive thinks about Brent Butterworth´s measurements at Soundstage Solo...

audiosciencereview.com

These headphones also tend to show the highest amount of variation between ear simulators in that "above feedback range" - "below a few kHz" range.
Ex : QC35 in Sean Olive's presentation for the HBK conference :

Rtings also tested the 5128 against their current ear simulator and SamV released some files in the wild.

I've been as a result very wary of over-interpreting some headphones' ear simulator results without additional information on their behaviour under several variables. For headphones such as the Bose 700, QC45 or Airpods Max, I've found ear simulator results to be a rather poor predictor of their actual on-head behaviour for me.

I could do some repeating experiments with and without the active function on.

 

[ADU]

Thank you for posting the additional plots above, Thomas. It looks like you're getting some very good results! I just had one or two quick questions though about how the FR plots of the other individuals are being compensated...

Did you use the same correction curve based on the in-ear response of your speakers measured in your own ears for all of the curves? Or did you also try to take in-ear measurements of the speakers for each individual, and then use their individual responses to the speakers to compensate just their own headphone measurements? (Hopefully that's not too confusing.)

 

[Thomas_A]

Thank you for posting the additional plots above, Thomas. It looks like you're getting some very good results! I just had one or two quick questions though about how the FR plots of the other individuals are being compensated...

Did you use the same correction curve based on the in-ear response of your speakers measured in your own ears for all of the curves? Or did you also try to take in-ear measurements of the speakers for each individual, and then use their individual responses to the speakers to compensate just their own headphone measurements? (Hopefully that's not too confusing.)

Hi,

no I just made one single measurement on me at LP. There are some many variables, I am not sure what more I can get out of that.

One more female included today so below are two males and two females in the panel (one additional male for QC25).

HD650: The open HD650 gives the most consistent results and is quite neutral, but is a bit "fat" in the upper bass and lacks lowest bass.
DT150/100P pads: As with HD650 quite neutral, but varies in bass depending on individuals. Needs careful fitting to get good bass.
DT770 Pro: Way too much treble and as with DT150, bass is dependent on individual and fit.
Bose QC25: A bit bass-heavy and fine up to 1 kHz. It's a mess between 1-4 kHz though, and no consistency between individuals/fit.

HD650, four individuals:

DT150_100P, four individuals:

DT770 Pro, four individuals:

Bose QC25, five individuals

 

[ADU]

Hi,

no I just made one single measurement on me at LP. There are some many variables, I am not sure what more I can get out of that.

Understood. It's interesting to see the variations in response for the different individuals. That could be done just as easily with the raw measurements though, without applying any sort of correction or compensation curve.

If you really want to see how a headphone deviates from the response of your neutral loudspeakers, then you'd have to do the in-ear measurements of the speakers on each individual as well. So you can use their own custom HRTF curve to correct or compensate the raw headphone measurements made on their head... It's more work. But that's really how compensation is designed to work.

And if it's done right, and the measurements of both the speakers and headphones are done in a consistent and highly repeatable manner on each individual, then the compensated plots for each headphone should look about the same, regardless of who it's measured on.

Calibrating each individual's measurements to the same reference stimulus or sound source (as opposed to the same curve), as I've described above, should effectively make all of the measurements platform- or rig-independent. So you can't really tell who or what system was used for the measurement.

 

[MayaTlab]

If you really want to see how a headphone deviates from the response of your neutral loudspeakers though, then you'd have to do the in-ear measurements of the speakers on each individual as well. So you can use their own custom HRTF curve to correct or compensate the raw headphone measurements made on that individual's head. It's more work to do. But that's really how compensation is designed to work.

One limitation here may come from the mics. I have no experience with the ones @Thomas_A used, but for measuring headphones, in-concha mics may introduce errors as low as 3-5kHz : https://audiosciencereview.com/foru...ted-harman-oe-curve-at-home.28130/post-975888

Some of the strong nulls you see above 5kHz for example could be a product of the mics, not necessarily representative of the headphones' actual behaviour on that individual.

Also, from what I understand, HRTFs usually are measured with mics that are at least at the ear canal entrance point. What you'd also want to measure is less one's HRTF, which I guess requires specialised facilities to measure, but rather "decent loudspeakers in a decent room", to "do a Harman".

What I do find most interesting here is the inclusion of the QC25 in the comparison and its behaviour in the 1-4kHz or so range. Most articles I've read on the subject of variance among individuals included mostly classic open or closed back passive designs like the HD650 or DT770 here (and @Thomas_A's results seem to follow the spirit of these articles' findings for these headphones), or stopped at 1kHz (ex : Harman's article on leakage).

 

[Thomas_A]

Perhaps OT but is it shown that the variable preference for bass is purely a preference or due to individual amounts of leakage?

Regarding >4-5 kHz response using concha mics, I’ll be back with some subjective scores. IMO, listening to sweeps and fixed tones is crucial if you want to do individual EQ.

 

[solderdude]

For QC25:

Below 800Hz the response is always correct due to noise cancelling (MFB basically)
From 1kHz to 5kHz the feedback mic, due to its placement and wavelengths as well as still 'partially' giving (incorrect) feedback it can go wonky in that part of the FR range.

 

[MayaTlab]

Perhaps OT but is it shown that the variable preference for bass is purely a preference or due to individual amounts of leakage?

For the over-ears, Harman took care to ensure that the headphones they used to test their target against alternative curves had a decent match between ear simulator and humans and didn't vary much between the latter. They mostly used large, over-ears open dynamics (K712, HD518, HD800), which have been shown to have low variation across listeners either in Harman's own article on leakage or from Rtings' data. In the 2013 article they also used an LCD-2.

For the in-ears they used a modded IEM with a MEMS mic inside the port. They equalised the response below 500Hz or so If I properly recall to get a consistent response below that frequency across listeners.

 

[MayaTlab]

For QC25:

Below 800Hz the response is always correct due to noise cancelling (MFB basically)
From 1kHz to 5kHz the feedback mic, due to its placement and wavelengths as well as still 'partially' giving (incorrect) feedback it can go wonky in that part of the FR range.

I don't think that we can be fully conclusive about the cause of these discrepancies, but I think that the feedback mechanism can be for the most part excluded as all Bose ANC headphones I've tested under pad compression show a similar behaviour in that range whether they're used with ANC ON or in passive, wired mode. Ex, QC45 :

As a comparison, here's how the HD650 behaves under pad compression :

Pad compression here is merely used as a proxy to test for several variables at the same time, in itself it isn't necessarily the variable at play hen testing across different users, but one of the variables tested might be involved in both cases.

What I have observed, however, is that some ANC headphones will react more strongly to a similar amount of pad compression when ANC is enabled in the range right above where the feedback range stops, but that isn't the case for Bose headphones, which seem to be purposefully designed to have a "null" point, insensitive to pad compression, right above where the feedback range stops : https://audiosciencereview.com/foru...se-use-of-the-harman-curve.29633/post-1087051

With a similar amount of pad compression (QC45, Sony H910N, Mark Levinson 5909) :

You may note that the Bose and Sony seem to share a similar behaviour under pad compression, with a fairly linear behaviour below a "null" point located around 1-2kHz and a very non-linear behaviour above. The Mark Levinson behaves under pad compression more like a traditional Harman closed back upon which an ANC system has been installed.

 

[solderdude]

Consider that in all cases though there is a clear 'unwanted' effect above the feedback range to 2-3 octaves above that point.
With the Bose it is higher up and happens to be at the same point where pad compression (pinna deformation ?) also happens to be.

Due to time delay (driver-FB mic distance) above a certain range phase shifts will occur. It looks like Bose has got this down a bit better though.
Bose NC in the QC25 and 35-II I measured is clearly better executed than the other NC headphones I encountered.

The ML5909 is not reacting as good as I hoped (certainly at that price point).

 

[MayaTlab]

Consider that in all cases though there is a clear 'unwanted' effect above the feedback range to 2-3 octaves above that point.
With the Bose it is higher up and happens to be at the same point where pad compression (pinna deformation ?) also happens to be.

The QC series leaves a lot of room for my ear lobes. I may have more on this later on, but I believe that the QC45's behaviour can be reproduced on a (boo evil !) flat plate (so without pinna).

My guess here is that someone knowledgeable would come here and start talking about acoustic impedance / porting yadiyada. I don't understand Bose's patents on the subject of ports / vent but they may provide some clues.

 

[Thomas_A]

With respect to pad compression, the clamp force of the HD650 was the strongest of the four. Depending on head size (which varied among listeners) I believe much of the overall different bass results was due to different pad compression. Even the QC25 showed differences 20 Hz to 1 kHz between subjects.