• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. 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!

10 Stages of Reading Headphone Measurements (by Resolve)

luft262

Senior Member
Joined
Mar 25, 2021
Messages
465
Likes
236
Location
Phoenix

This was a pretty fun list by Resolve. I think it may be a little more reflective of the phases he's personally been through in his own thinking on headphone measurements though, than some others. :) Though I can certainly identify with alot of what he's talkin about here. And I suspect some others here may as well. (Hence the reason for this post.)

There are some things I agree with in this, and others I don't though.

And I hope that stage 11 is realizing that we need more in-ear frequency response data on good-sounding loudspeakers in a room for comparison with the in-ear data of headphones. Preferably measured on the exact same anthropomorphic rigs (with a mannikin) that are used for the headphone FR measurements.side
Since I brought up some of the shortcomings in the 711 system, I should also mention that in some recent studies, Harman researchers also uncovered some possible issues with the accuracy of the HBK's 5128 mannikin in terms of its potential leakage/interfacing effects with certain headphones. The study is discussed in the latter part of this video for a recent HBK conference. I've pre-cued the video to where that particular info begins. The rest is a summary of Harman's other work and findings related to headphones (so feel free to back it up to also give that a listen, if you wish.)


I believe there was some additional discussion on this subject in a couple topics here on ASR as well. This is one of them, but I think there may have been at least one other as well. (This is goin back several months now, so my memory is a bit fuzzy.)

Side question, but how do you get the "addicted to fun and learning" tag under your's and others names?
 
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
Side question, but how do you get the "addicted to fun and learning" tag under your's and others names?

I assume it's based on your number of posts, luft262. :)
 
Last edited:
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
I think @ADU is referring to this type of measurement. This was how Dr Olive's (Harman) headphone target curve got started. Picture is from Tyll Hertsens' trip to Harman with his HATS (head and torso simulator).

index.php

In addition to the torso being missing, it looks like the speakers might also be positioned too close to the wall in the above setup for good stereo imaging. Alot of average Joe's will position their speakers too close to the walls in their homes as well. I don't know if that was the reasoning though behind this arrangement. Or if it might have been related in some way to achieving what Harman might have felt was a better calibration of the speakers and sound in the room, to meet some of their own internal specs.

The orientation of the dummy and speakers probably needs to be close to an equilateral triangle though, as shown here. (Though some prefer the speakers to be a little closer together for a better phantom center.) And I think the toeing in is probably also correct, though that may vary by model and driver design.

The room is not completely damped btw. And that is by design. Because it is intended to more or less mimic the semi-reflective characteristics in a typical home listening space. The damping and reflectance are adjustable though in Harman's reference listening room.
 
Last edited:
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
Do you have the link? Thanks!

I'm not that adept with Twitter, but I believe you can find most of Dr. Olive's comments in this thread. You just need to keep hitting the "Show replies", and "Show more replies" commands to see them all.

You may find some others on Dr. Olive's Twitter page though, if you scroll down a bit.
 
Last edited:

Earfonia

Active Member
Joined
Feb 21, 2019
Messages
278
Likes
398
Location
Singapore
I'm not that adept with Twitter, but I believe you can find most of Dr. Olive's comments in this thread. You just need to keep hitting the "Show replies", and "Show more replies" commands to see them all.

You may find some others on Dr. Olive's Twitter page though, if you scroll down a bit.

Thank you!

2022-02-18_115023.png


I'm interested with Dr. Sean's question: Are You really hearing the HARMAN Target?

We can start with this group / community. If we have let's say same model of an IEM, we can create an EQ profile for the IEM to sound very close to Harman Target using Equalizer APO. Listen to it and give our thought about the tonality.

I did that to test my own target curve. To me untested target curve is not worth to be a target. IMHO we should balance theory discussion with real listening test.
 

luft262

Senior Member
Joined
Mar 25, 2021
Messages
465
Likes
236
Location
Phoenix
Not a problem. If you have other questions about anything above or in the video in the first post, feel free to ask, and I'll try to cobble together an opinion or two that will hopefully make a little sense. :) (Maybe Resolve will also drop by and give his take on some of this as well, which would be interesting).

One thing I should probably have mentioned is that the spinorama speaker data gathered by setups like Amir's Klippel system is still an important component in all this, even if you're doing the other in-ear measurements with a head and torso simulator rig. Because without the spinorama data or anechoic measurements, there's no real way of knowing if the speaker being used for these kinds of in-ear tests has a truly flat direct response, and a smooth linear-ish off-axis response. Because that is not the sort of info you can really get by looking at a speaker's in-ear response.

The torso, head, and ears on such headphone measurement rigs will all distort the frequencies of the speaker's in-ear response in such a way that a neutral response would no longer be represented by just a flat line, like it is for the direct response or listening window in a spinorama plot. So the in-ear data is more or less unusable as a tool for assessing the speaker's accuracy and neutrality in a normal listening space.

So both the in-ear, and the out-of-ear or anechoic measurements have their places in this. And are critical to understanding whether the speaker and headphones with a similar response are approximating something which could reasonably be called "neutral".
Is there something like the Harmon Curve for stereo speaker response in a regular room? Is there a generally accepted target curve for a stereo speaker system or would that also be the same Harmon Curve? Thanks.
 
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
Thank you!

View attachment 187382

I'm interested with Dr. Sean's question: Are You really hearing the HARMAN Target?

We can start with this group / community. If we have let's say same model of an IEM, we can create an EQ profile for the IEM to sound very close to Harman Target using Equalizer APO. Listen to it and give our thought about the tonality.

I did that to test my own target curve. To me untested target curve is not worth to be a target. IMHO we should balance theory discussion with real listening test.

I am less knowledgeable re IEMs than you probably are, Earfonia. But if Harman had released a version of their headphone target that was compensated with an accurate diffuse field response curve, then it probably would've been much easier to do a more reasonable job of translating it to some other platforms. Especially those where the DF HRTF responses had already been measured and were known. Raw response curves are much less useful for this kind of thing, and going back and forth between different measurement rigs and setups, imho.

Since the Harman target was designed primarily for Harman's own internal use in designing and improving the sound on their own headphones, and releasing a DF version of their target would likely have helped many of the competitors, this was probably not very high on their list of priorities. It probably would have made the implementation and acceptance of their rating system much smoother and easier though. So it was a significant shortcoming in their approach.

Unfortunately, in some of their early research, they apparently saw the diffuse field as a dead end, perhaps because they were looking at it primarily from a tonal balance standpoint (and finding some obvious shortcomings in that respect). And may not have stopped to think that it could've had some other benefits in terms of translating their results for use on other systems and platforms,... including some of the other rigs they were using for their own research, like the newer HBK 5128 system. That is probably water under the bridge though now.

This is a bit OT, but most of Resolve's most recent livestream was dedicated to the subject of IEMs, btw, Earfonia. In case you may be interested in giving that a listen...

 
Last edited:
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
Is there something like the Harmon Curve for stereo speaker response in a regular room? Is there a generally accepted target curve for a stereo speaker system or would that also be the same Harmon Curve? Thanks.

If you are asking whether or not there is a generally accepted curve that is ideal for most rooms, then the answer is probably no. Because every room is different, with its own unique set of acoustic properties.

There are certain characteristics that you should look for though in a good speaker. And those are a flat direct or on-axis response (though some also prefer to use the listening window for this), and a reasonably smooth directivity and off-axis response. You need all of these characteristics to achieve what Floyd Toole describes as a "neutral response". (See videos below for more on this.)

Good extension into the sub-bass and higher frequencies would also be another important characteristic imho. (Though the former can be achieved with the addition of a good sub-woofer.)

These would all be based on "out-of-the-ear" measurements, rather than in-ear measurements made with a dummy btw.


 
Last edited:

Earfonia

Active Member
Joined
Feb 21, 2019
Messages
278
Likes
398
Location
Singapore
I am less knowledgeable re IEMs than you probably are, Earfonia. But if Harman had released a version of their headphone target that was compensated with an accurate diffuse field response curve, then it probably would've been much easier to do a more reasonable job of translating it to some other platforms. Especially those where the DF HRTF responses had already been measured and were known. Raw response curves are much less useful for this kind of thing, and going back and forth between different measurement rigs and setups, imho.

On the video you posted earlier:

As well as from the following slides (same content):

Harman used GRAS 45CA for the Harman target measurement.

My measurement equipment is not GRAS 45CA but calibrated to match GRAS 45CA. But only for IEM measurement. With the right insertion depth, the IEM measurement result should be pretty close to GRAS 45CA. So it can be used to EQ an IEM then measure it to achieve Harman Target Curve.

On the contrary, in my opinion raw response curve is the right way to present headphones and IEMs frequency response. With the lack of consensus for universal target curve there is no logical reason to compensate raw frequency response. Compensation is adding 'interpretation' to the result that might be totally different than they way we interpret the raw response. When measurement presented raw it is still useful enough to apply different 'interpretations' / target curve to it. But once it is compensated with target curve that is different to our preferred target curve, the compensated curve is totally useless.
 
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
On the video you posted earlier:

As well as from the following slides (same content):

Harman used GRAS 45CA for the Harman target measurement.

My measurement equipment is not GRAS 45CA but calibrated to match GRAS 45CA. But only for IEM measurement. With the right insertion depth, the IEM measurement result should be pretty close to GRAS 45CA. So it can be used to EQ an IEM then measure it to achieve Harman Target Curve.

On the contrary, in my opinion raw response curve is the right way to present headphones and IEMs frequency response. With the lack of consensus for universal target curve there is no logical reason to compensate raw frequency response. Compensation is adding 'interpretation' to the result that might be totally different than they way we interpret the raw response. When measurement presented raw it is still useful enough to apply different 'interpretations' / target curve to it. But once it is compensated with target curve that is different to our preferred target curve, the compensated curve is totally useless.

Yes. I can understand where you are coming from on this, Earfonia. And why using the diffuse field seems like the wrong approach. And it is clear to me from your other posts that you have spent a good bit of time examining the Harman research. I am looking at this more from the standpoint of over-ear headphones though. So I may be looking at the situation and data a bit differently. And from my perspective, "interpretation" is really sort of the whole point of using the diffuse field compensation.

As we touched on earlier in this topic, each different manufacturer's measurement rig has it's own somewhat unique and different HRTF. So there should be some form of interpretation, translation, adaptation, calibration, compensation, correction, or adjustment in the Harman target (or any target response curve, for that matter) for it to really make sense across the board on other measurement platforms than the one (or ones) that were used to create and develop it. And using compensation with a diffuse field HRTF would be one of the possible methods of accomplishing this.

It may not necessarily be the best or only method of doing this. But because most HATS systems will generally include a diffuse field HRTF compensation curve for their rigs, it is (or rather, would have been) one of the easier and more readily available methods of translating, adapting or interpreting a target response curve like the Harman over-ear target for better use across the different measurement platforms.

Using diffuse field compensation (or at the very least, supplying the necessary DF data from their measurement gear to make this possible) would've made Harman's target much more platform-independent, and more readily adaptable and usable by other measurement systems than just the proprietary setups that were being used for their own in-house research.

I suppose it is also possible that supplying this type of DF info could've revealed too much info about their research setup though, potentially making it too easy for their competitors and other manufacturers to replicate and duplicate their in-house results. And maybe that's why they were hesitant, or unwilling, or simply unable to release this kind of measurement data for their particular gear into the public domain.

It is also possible that they regarded the differences between the different 711 measurement systems as being too insignificant to warrant the extra trouble of including this type of adaptive measurement information with their target. And regarded the raw data as being adequate or sufficient for most of their purposes.
 
Last edited:

Earfonia

Active Member
Joined
Feb 21, 2019
Messages
278
Likes
398
Location
Singapore
Just to clarify, what I mean by compensation is target curve compensation like Harman target, etc. It is completely different than microphone calibration of the IEC coupler / ear simulator. I believe IEC coupler's microphone should be calibrated to a certain standard to make measurement results from different IEC coupler to be more compatible. For example my clone coupler microphone is calibrated to GRAS 45CA.

And what I mean by RAW measurement result is measurement result from a calibrated IEC coupler. So mic calibration must always be factored in.
Compensation is using a target curve to offset / compensate the RAW measurement result.

Just to clarify the difference between calibration and compensation.
 
  • Like
Reactions: ADU
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
Just to clarify, what I mean by compensation is target curve compensation like Harman target, etc. It is completely different than microphone calibration of the IEC coupler / ear simulator. I believe IEC coupler's microphone should be calibrated to a certain standard to make measurement results from different IEC coupler to be more compatible. For example my clone coupler microphone is calibrated to GRAS 45CA.

And what I mean by RAW measurement result is measurement result from a calibrated IEC coupler. So mic calibration must always be factored in.
Compensation is using a target curve to offset / compensate the RAW measurement result.

Just to clarify the difference between calibration and compensation.

We are on the same page on this, Earfonia.
 
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
Is there something like the Harmon Curve for stereo speaker response in a regular room? Is there a generally accepted target curve for a stereo speaker system or would that also be the same Harmon Curve? Thanks.

You could also try to compute an average in-room (or sound power!) response curve from the spinorama measurements of several different speakers with a neutral response, luft262. And this might give you a little better idea of what a "typical" in-room response might possibly look like.

The typical response of a neutral loudspeaker can vary though depending on the specific features or characteristics you're indexing for. The graph below shows a couple different examples of what I mean by this...

5BESTEXTENDED 10V-SHAPED SP SEPARATED.jpg


I sampled the sound power responses (which is different than in-room responses) of 25 different loudspeakers with a relatively flat direct/on-axis response and good bass extension in Pierre's spinorama database. And the curves above represent the average sound power of two different groupings or subsets of those 25 speakers.

The first curve in green is an average of the 5 best-extended speakers in the sub-bass. And the 2nd curve in pink is an average of 10 speakers with what I call a more "V-shaped" response. If I overlay the two curves, then maybe you can see the differences between them a little better...

5BESTEXTENDED 10V-SHAPED SP OVERLAID.jpg


The differences are somewhat subtle in the midrange and treble frequencies.

Both sets of speakers have a slight dip though in the upper mids at around 2 to 2.5 kHz where the midrange drivers and tweeters cross over, because there's a slight reduction in their dispersion (or increase in directivity) in that range. The dip is a bit more obvious though on the speakers with a V-shaped response. Perhaps because they do not have quite as good cross-over designs on average.

The average of 10 V-shaped speakers also does not extend as deeply into the lower frequencies as the 5 best-extended speakers. Because the more V-shaped curve is based on a larger sampling of speakers, with poorer LF extension overall. The average of the 5 best-extended speakers in green might be a little closer to the type of SP response you'd get if you paired a speaker with good midrange and treble response with a sub-woofer though.

There are other groups of speakers which can have a more linear, or a more "L-shaped" SP response than the above examples as well.
 
Last edited:

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,739
Likes
16,166
The problem is that humans don't really perceive the tonality of loudspeaker listening position curves recorded by non gated omnidirectional microphones but above transition region rather the direct sound which remains flatish, which is one of the reasons that generic loudspeaker targets at the listening position don't work. The universal target for loudpeakers is flat direct sound and smooth directivity, the listening position curve then depends on the directivity, listening distance and absorptivity of the room, more can be read about it in the first link in my signature.
 
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
The problem is that humans don't really perceive the tonality of loudspeaker listening position curves recorded by non gated omnidirectional microphones but above transition region rather the direct sound which remains flatish, which is one of the reasons that generic loudspeaker targets at the listening position don't work. The universal target for loudpeakers is flat direct sound and smooth directivity, the listening position curve then depends on the directivity, listening distance and absorptivity of the room, more can be read about it in the first link in my signature.

Thank you for mentioning this, thewas.

I thought I'd already mentioned this here, but guess it was a different topic... But I generally agree with Floyd Toole's contention that it's not a good idea to use the in-room response of a speaker for EQ purposes. Because doing that can also potentially compromise a speaker's direct/on-axis performance. I think Dr. Toole makes this point pretty strongly in both the videos I posted earlier (and probably also in your link). And that was part of my reason for posting those. Generally speaking, I think you're better off using EQ to fix errors or imbalances in a speaker's direct/on-axis response, which is something that the anechoic and spinorama measurements are good for.

People like to do their own measurements though. And that can also give you at least some info about the effects that any tweaks to the direct sound may be having on your speaker's in-room response, if nothing else. So there is a place for both things imo.
 
Last edited:
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
Unfortunately, I don't think there's really any precise equivalent to a speaker's direct/on-axis response or directivity in a pair of headphones. At least not directly. If there were, that would certainly make things alot easier. There are some other characteristics that we can compare though.

We can, for example, do a direct comparison between the raw in-ear response of a headphone, and raw in-ear measurements of a speaker's steady-state or in-room response. And get some ideas about the similarities from that.

The significance of this relationship was actually one of the principal findings of Harman's headphone research, even though it has taken somewhat of a backseat to the other discussions of targets, preference testing, and so forth (which incidentally helped to confirm this relationship).
 
Last edited:

TurtlePaul

Addicted to Fun and Learning
Joined
May 19, 2020
Messages
631
Likes
1,027
Location
New York
Harman is like 25 years ahead of you guys. The target for loudspeakers os an in-room response which is flat but tilts down by just under 1 dB per octave, plus a little bass boost.

1645414330240.png
 
OP
A

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,584
Likes
1,086
Harman is like 25 years ahead of you guys. The target for loudspeakers os an in-room response which is flat but tilts down by just under 1 dB per octave, plus a little bass boost.

If you really believe this, TurtlePaul, then perhaps you'd better take another look at some of the Toole videos and links posted above. ;) Some new data has been accumulated though in the last 25 years on the in-room and dispersive characteristics of loudspeakers, thanks in part to websites like this one.

I'm going to turn my thoughts back to the subject of headphone measurements though (at least for the moment). And some of the other potential relationships they could have with the measurements of loudspeakers imho.

These are the same two average sound power loudspeaker curves shown in one of my previous posts above. But with one minor change or addition...

5BESTEXTENDED 10V-SHAPED SP PLUS 5128DF SEPARATED.jpg


Both of the curves have now been combined with the diffuse field HRTF response of HBK's new 5128 HATS measurement rig.

The diffuse field response of the 5128 is represented by the dashed purple curve on this ASR graph of the Sennheiser HD-650 headphone...

index.php


It's the curve with the brightest tilt or response on the graph. And is noticeably brighter in the treble than both Amir's recalibrated version of the Harman curve for the 5128, and also the measured in-ear response of the HD-650 (which is considered by some audiophiles to be a fairly neutral-sounding headphone in the midrange and treble).

To a casual observer, the 5128 DF curve looks way too bright to really be of much use for analyzing the in-ear responses of headphones (or anything else for that matter). What the purple DF curve represents though is the HBK 5128's HRTF response to a spectrally flat sound field that is essentially equal (or diffused) in all directions.

When you're listening to speakers in a room in your home, sound is also coming at your ears from all different directions, like in a diffuse sound field. The volume and timbre (and arrival time) of the sound is not the same though from all directions. Because both the directivity or dispersion characteristics of the speakers, and the acoustics of the room reflecting the sound back to your ears will change the timbral balance or composition of the speakers so they have a darker overall tilt than they would if you were listening to just the speaker's direct/on-axis response in a more heavily damped or non-reflective (aka echo-free or anechoic) space. This is because most speakers disperse their sound more broadly at lower frequencies than at the higher frequencies. So the room tends to reflect more LF back to the listener than HF. And the net result is an in-room response that's tilted more in favor of the bass, and less in favor of the treble (like my homey TurtlePaul just mentioned above).

So what does this have to do with my headphones, or the price of beans in Chile?... Well, it turns out that if you take something like a HATS measurement rig's, or any person's in-ear response to a spectrally flat diffuse sound field, and then combine that with a speaker's spectrally-tilted sound power (which is just a summary of it's diffuse response in all directions in a room), what you seem to get is something that looks awfully close to the speaker's probable in-ear response in an average domestic listening space.

It kinda makes sense when you think about it, because what the normal DF response of a measurement rig is generally missing is that darker "tilt" that you get when listening to speakers in a semi-reflective room. And the sound power of a neutral loudspeaker seems to fill in that missing timbral information quite nicely. So combining the two should give you something fairly close to an approximation of a speaker's in-ear response, if you were listening to them in a typical semi-reflective room (which is one of the models Harman uses in its spinorama calculations).

How that all works is still a bit of a mystery to me, since I'm not much of a math or engineering genius. But it seems to be borne out, at least implicitly, by much of Harman's data and research. And also the independent measurements I've looked at on both headphones and speakers, including the graphs in Pierre's spinorama database, and also raw and compensated headphone measurements by ASR and other headphone graphers/reviewers (such as Oratory1990, Crin, Resolve, Inner Fidelity, Rtings, the SoundGuys, and so forth).

This is not something which can really be proven though until there are more actual in-ear measurements of speakers to compare with sound power + DF HRTF results, like the ones that I've posted above.

These are the two average sound power curves modified with the 5128 DF in-ear HRTF curve, overlaid for a little easier comparison btw...

5BESTEXTENDED 10V-SHAPED SP PLUS 5128DF OVERLAID.jpg


Note the slightly more pronounced dip in the mids on the pink curve (which represents the V-shaped speakers with some cross-over/directivity issues).

Although these curves have a fairly "Harman-ish" look to them, there are a few differences between them and the over-ear headphone target Harman developed for it's own GRAS and KEMAR-based measurements, particularly in the treble. So these curves are mostly useful for comparison just with the other HBK 5128 measurements made by ASR, or by Jude at Head-Fi. And by one or two other sites that are starting to use HBK 5128 systems for their reviews (like some of the recent 5128 headphone plots by the Sound Guys).
 
Last edited:
Top Bottom