• 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!

Resolve's B&K 5128 Headphone Target - you can try the EQ's.....

Status
Not open for further replies.

MayaTlab

Addicted to Fun and Learning
Joined
Aug 15, 2020
Messages
959
Likes
1,610
I, from ignorance, don't quite understand the benefits of using this B&K measurement rig and everyone turning to it suddenly.

I've the idea that it will present a FR closer to what we hear (never the same, because of our personal HRTF issue), but anyway, those benefits are present after 9-10 kHz, an area whose accuracy may not serve us too much to make better EQ decisions anyway, no?

The impedance of the 5128 system is closer to the average human population than the 711, and this will have an effect throughout the entire spectrum : https://audiosciencereview.com/foru...target-you-can-try-the-eqs.43209/post-1569672
There's otherwise been a more stringent effort to get the geometry closer to the average ear canal dimension.
It also unfortunately is quite prone to produce rocking modes / wiggles at lower frequencies that are not apparent in the few in-situ measurements that I've seen so far, but if it is like with other pinnae it can be remedied with putty to some extent.

I'd like to see more in-situ measurements of IEMs (preferably closer to DRP), comparing 711 and 5128 couplers to real humans, but so far all of the in situ measurements that I've seen tend to support the idea that, at least below 1kHz where in situ IEM measurements are easier to perform and interpret than above, the 5128 is indeed more representative of how IEMs behave in actual humans (minus the rocking modes).

So, what's the point? For R&D and production environments I could understand its added precision,

Some would say that difficulties with positioning the IEMs in a repeatable fashion in its pinna (and that goes as well for 711 systems using a pinna) are likely to make such systems less precise than simply using a 711 coupler with a canal extension. I think that you rather meant accuracy ?

I think that it would be quite interesting to see a system that mimics the transfer impedance of the 5128 in a format closer to 711 couplers + canal extension, if possible.
 

MayaTlab

Addicted to Fun and Learning
Joined
Aug 15, 2020
Messages
959
Likes
1,610
And your justification for 'slope not shelf' is also flawed.

Headphones.com's argument for using slopes instead of shelves isn't only related to the 2013 article, and you know it.

The test conducted by Harman in which the former was preferred was done using an Audeze LCD2, which is an acoustic outlier in that it has an almost completely airtight front volume (so much so people have complained about a suction effect when taking them off, which I believe Audeze rectified by introducing some front venting in other later models). This all comes back to your old friend bass slam/impact. As I've told you before, all else equal perceived bass impact likely correlates with degree of front volume seal. A bass shelf predominantly accentuates low and sub-bass, which also correlates with bass impact. When you combine the two effects, some/many are likely to find this high bass impact too overpowering, and so with these (very particular) headphones prefer a slope, which does not predominantly emphasize the low/sub-bass. This would also explain why DF was not rated that badly with these headphones, as they naturally provide some bass impact even with flat DF-bass. The other test in this same study (which funnily enough you rarely if ever mention) used the Sennheiser HD518, with a much more common non-airtight front volume design, with which shelf was rated considerably higher than DF and with a higher rating than both on the LCD2. Again, the difference in results between the headphones can be explained by front seal differences, with the bass shelf providing the perceived bass impact needed without an airtight front volume, the latter few (if any modern?) headphones have, and so test results involving such outlier headphones should not be used to base targets around.

Notwithstanding the double standard, as you don't seem to require from your own ideas the same degree of vetting by listening tests as you'd like to see from others (where's the evidence for what's bolded ?), do you realise that the frequency response of the LCD-2 Harman equalised to the slope-based target (RR1_G) turns out to be closer to any of the shelf-based OE Harman targets that have been produced over time, than the shelf-based (RR_G) target anyway ? Have you actually read that article or only the blog posts, pdf or powerpoint documents on its subject ?

Perhaps let's not follow excessively simplistic interpretations then, there are quite a few variables other than the one you've mentioned that could be hypothesised to have caused what you observed.

In the end I'd prefer to see listening tests to support the DF HRTF + slope (and / or other preferential adjustments) approach, but in the meantime I'd also prefer to see counter-arguments come from a genuine place of interest instead of a pathological antipathy towards Headphones.com, which by now probably comprises most of your interventions on ASR.
 

markanini

Major Contributor
Joined
Feb 15, 2019
Messages
1,816
Likes
1,892
Location
Scania
A major fundamental issue, mentioned by Resolve himself in Twitter, as well as Crinacle in his video - to the point where he advises against direct treble comparisons of BK5128 measurements, is the difficulty to achieve consistent insertion on BK5128. On IEC711 this is trivial. I haven't heard anyone with the 5128 at hand finding a solution to this issue. Until then this will be a massive hurdle for reaping the perks of the system.

This has to beg the question, what is the knowledge and/or honesty of anyone that doesn't acknowledge this? If your experience was like mine it's a touchy subject that's has the hallmarks of having some hidden incentive. In private discord servers I get dragged into unsolicited debates where BK5128 measurements are falsely presented as reliable data, by people that should know better. I've never seen such overzealous importance put on measurements systems on traditionally subjectivists channel before BK5128 adoption. That should tell you something.
 
Last edited:

MayaTlab

Addicted to Fun and Learning
Joined
Aug 15, 2020
Messages
959
Likes
1,610
A major fundamental issue, mentioned by Resolve himself in Twitter, as well as Crinacle in his video - to the point where he advises against direct treble comparisons of BK5128 measurements, is the difficulty to achieve consistent insertion on BK5128. On IEC711 this is trivial.

You mean per-individual / operator, or across measurement databases ? Per IEM or across several IEMs within the same database ? Trivial with an ear canal extension, or even with the pinna (anthropometric) ?

I don't think that you should be too obsessive with how IEMs measure in the treble range regardless of the type of coupler used anyway.
 

markanini

Major Contributor
Joined
Feb 15, 2019
Messages
1,816
Likes
1,892
Location
Scania
I don't think that you should be too obsessive with how IEMs measure in the treble range regardless of the type of coupler used anyway.
Completely agree. Only specific contexts warrants that, like calculating deltas between different measurements systems. Absolutely not for evaluating IEMs.

You mean per-individual / operator, or across measurement databases ? Per IEM or across several IEMs within the same database ? Trivial with an ear canal extension, or even with the pinna (anthropometric) ?
The fact that the coupler and pinna is highly integrated, physically preventing the operator from choosing a desired insertion depth. AFAIK no one has committed to making pinnaless measurements on BK5128. I don't know if pinnaless measurements are officially supported in the first place, it seems everyone with the rig at hand is apprehensive about it.
 

Resolve

Active Member
Reviewer
Joined
Jan 20, 2021
Messages
212
Likes
531
So, some of the speculation here that nobody else uses this same calculation is actually false, since it's the same one Crinacle uses as well. And I repeat, if anyone has a different calculation for DFHRTF I'm happy to test that out. My understanding is that actually doing this manually may yield a different result than as a calculation, which is what it is. IMO if there's any issue to be taken with the fine-grained elements of this target (and again it's what's being used for the largest public 5128 graph information so far), then determining what the DFHRTF should be is the locus of the debate. The idea that this is somehow our specific editorialization on how headphones should measure to make headphones look better is unfounded nonsense.

We've already learned a lot by using the 5128, in particular when it comes to IEMs like with the Blessing 1 and 3 bass level revelation, and comparing data between the two 'heads' (GRAS and B&K) is already interesting. The previous measurement paradigm essentially ignored HpTF effects - at least when it came to the way people commonly evaluated them - and I have yet to come across a compelling reason why we as a community shouldn't want that data. So far what I'm seeing is some version of "I don't understand it, so I don't want it". While I can empathize with that, it's not a good enough reason to not care about this information.

Lastly, as it relates to Harman - yeah I'm looking forward to them doing research on the 5128. Maybe we can convince them to make it public...
 
OP
Robbo99999

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
7,062
Likes
6,934
Location
UK
Doing a critique exercise, I think all of this it's a byproduct of the average reviewer mindset (and I'm not saying that as a personal attack on Resolve, or that it's inherently a bad thing), it's just their minds are constantly focused looking for new and cutting edge information about the audio industry, and that's fine, but the mistake in my opinion is to believe that everything is determined by a "higher spec" and lack common sense in the most elementary questions.

I.E.: If something that plays a critical role in a specific function isn't used and endorsed by experts in audio research yet (such as Sean Olive and the Harman team), then don't try to reinvent the wheel and pretend that you know more than the rest of them, especially when you're not an expert on the subject!
The good thing is you can always go back to the old trusty GRAS - it's a bit like when you try a new headphone EQ or a new headphone & you think it's the bees knees at first, but then you compare it to your long-standing EQ and you realise you'd already hit the nail on the head. But, but, I don't have a problem with people striving to find new target curves on a different measuring kit, and it helps if there's some scientific rational behind it. It's interesting to try new options, and hell even if Resolve & Co ultimately find out that they can't improve on GRAS Harman then it's an endeavour & experience & they can graciously go back to GRAS Harman. That might be a bit galling after the financial outlay of the kit, and time spent on it, but if you have the right mindset you can put it down to the goal of just trying to further optimise headphone sound that's based on a mannequin measurement device. The only reason you can think Harman GRAS is the pinnacle of headphone sound (on average for most folks) is if it is indeed truly the best average target for most folks, and it's hard to accept that it's been totally nailed in terms of that average in all areas of the frequency response......but at the same time it's probably a big ask to better it, as it would mean needing a better starting point and also a better evaluation process, which is not easy. But I admire them wanting to try something different & I'm curious as to how they proceed & what they create, there's some knowledgeable people working together on it, so that's a good basis to begin with....it might take a while, and potentially they may themselves realise they can't come up with anything better, but perhaps they will.....and they will need to try to have some good self integrity to let themselves not be biased in their final comparisons if it doesn't quite live up to their expectations.
 

Resolve

Active Member
Reviewer
Joined
Jan 20, 2021
Messages
212
Likes
531
even if Resolve & Co ultimately find out that they can't improve on GRAS Harman then it's an endeavour & experience & they can graciously go back to GRAS Harman.
Yeah we've certainly used Harman before and can do so again. I will say, I think a more likely scenario is that they do new preference research with the 5128 or other HATS on the new ITU standard, and if that's similarly satisfying/comprehensive there's no reason we can't all just use that - apart from it not being publicly available. What I'd like to see from them - and I talked to Dr. Olive about this and he thought it was a reasonable idea - is a more explicit three point adjustment. I say more explicit because technically ear gain did get adjusted in their previous research. But all of this is of course wishful thinking, since the current line from Harman is that new research isn't being made public.
 
OP
Robbo99999

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
7,062
Likes
6,934
Location
UK
Yeah we've certainly used Harman before and can do so again. I will say, I think a more likely scenario is that they do new preference research with the 5128 or other HATS on the new ITU standard, and if that's similarly satisfying/comprehensive there's no reason we can't all just use that - apart from it not being publicly available. What I'd like to see from them - and I talked to Dr. Olive about this and he thought it was a reasonable idea - is a more explicit three point adjustment. I say more explicit because technically ear gain did get adjusted in their previous research. But all of this is of course wishful thinking, since the current line from Harman is that new research isn't being made public.
Yep, if they don't make it public, then it's just down to what you guys can create on it.....and if all else fails back to GRAS Harman, which really shouldn't have any shame attached. Afterall, the future of headphones may not be based on mannequin derived target curves but more on individualised HRTF and DSP, so that might be a future focus of any headphone reviewer, if it gets there.

EDIT: I hope they make it public if they do some extensive work on the 5128.
 
Last edited:

Resolve

Active Member
Reviewer
Joined
Jan 20, 2021
Messages
212
Likes
531
Afterall, the future of headphones may not be based on mannequin derived target curves but more on individualised HRTF and DSP
Which is one of the reasons Dr. Olive wants to focus a lot on the personalization aspect of TWS for example. I think it's a good idea, and talking about how variant or invariant a headphone is for HpTF effects is something we can already do now.
 

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,361
Likes
1,893
So, some of the speculation here that nobody else uses this same calculation is actually false, since it's the same one Crinacle uses as well.
Straw man. We're not talking about the DFHRTF calculation, we're talking about the target actually used to display and compensate measurements, and the default target Crinacle displays for the B&K on his graph tool is this:
graph-112.png

i.e. DFHRTF with a -8 dB (-0.8 dB/octave) slope and a +10 dB bass shelf, relative to DF. Here it is not relative to DF (not completely unsmoothed by default either):
graph-113.png

Compare with your target (dashed line) that has a -10 dB slope and no shelf:
TruthEar Zero Red 5128.jpg

Clearly these are considerably different. And if you're weakening your argument to say they have generally similar 'fine-grained' features, 1) that's pretty inconsequential, not only psychoacoustically (high-Q, low amplitude => less audible), but especially in comparison to the large coarse-grained differences and 2) that's not surprising considering Crinacle is sponsored by headphones.com. And yet despite this it looks like even he realizes the target sounds wrong (as many have said, thin/bright/shouty etc.) without a bass shelf (which funnily enough is also a tacit admission that his own Truthear Red he 'tuned' lacks bass, as they fall significantly short of his target down there stock, w/o resistor). Also, here's a bonus game of spot the odd one out:
FqkNSfuaYAA27xz

The idea that this is somehow our specific editorialization on how headphones should measure to make headphones look better is unfounded nonsense.
Another straw man, that's nice of you to give the first one a friend to play with. The pluralist nonsense you're actually planning in order to make headphones/IEMs look better is even worse.
So far what I'm seeing is some version of "I don't understand it, so I don't want it".
Ironic, considering you're clearly not understanding then rejecting what people have repeatedly been trying to tell you ever since you first got the 5128: raw data is of little use without proper processing, and in this case that means compensation to a target that has direct demonstrated correlation with preference in the form of controlled blind listening tests. This is paramount. Without this, flawed supposedly 'based in theory' target concoctions won't cut it.
Lastly, as it relates to Harman - yeah I'm looking forward to them doing research on the 5128.
They've done some research on the 5128 already, and produced an extrapolated target that you all at headphones.com have derided. Maybe before you poo-poo it, how about demonstrating that your target is any better with at the very least a lower mean predicted preference rating error than the 5.8 points that this 5128 target from Harman has in relation to the predicted ratings from the original GRAS measurements and Harman target as Sean has shown. Of course even that's a fairly low bar, and what you'd actually need to empirically substantiate the casually thrown around claim that the 5128 provides a more accurate representation of human headphone perception/preference than GRAS is a correlation between predicted and actual preference ratings of greater than the high 86% Harman achieved using their GRAS rig and target for it (as well as the high 85% virtualization correlation, and that was with the old KB007X too, before they moved to more anthropomorphic pinnae which would likely increase this correlation even higher). As for IEMs, Harman's preference correlation was 91% (= 0.91 for those who don't understand basic math), and for people Addicted to FUD and Moaning about Harman hilariously saying these correlations are not very good and desperately trying to pick holes in their research, there is also independent evidence the Harman target is preferred from companies like USound where Oratory works, who's stated Harman IE is preferred in blind tests to his own IE target (which coincidentally? the Truthear Red /w resistor follows suspiciously closely, which in turn follows Crin's supposedly-better-than-HarmanIE 5128 target) in non-noisy environments (which are the conditions audiophiles and reviewers judge sound quality under). Yet more recent independent blind tests corroborating Harman's in-ear research conducted by Danish scientists found listeners ranked tested IEM frequency responses (measured using the claimed more accurate 5128 for all the B&K fanatics) in non-noisy environments in the order predicted by their adherence to the Harman IE target as calculated by Dr Sean Olive's algorithm. All this Harman achieved with a supposedly 'innacurate' GRAS rig. The Harman Science Deniers refusing to accept the validity and utility of the in-ear research and predicted preference ratings based on the GRAS-developed Harman target don't have a leg to stand on. And what valid preference evidence do they have to offer for a 5128 target in return? Absolutely zilch.
 
Last edited:

MayaTlab

Addicted to Fun and Learning
Joined
Aug 15, 2020
Messages
959
Likes
1,610
Also, here's a bonus game of spot the odd one out:
FqkNSfuaYAA27xz

I'm surprised that you can even read this graph since it wasn't normalised at 500Hz. OMG Harman is massaging the data !
I'm also surprised since you like to nitpick that you didn't notice that Harman's digitisation work must have been a bit rushed as there are some differences between these targets that don't make a lot of sense given how they were designed. Not that it should matter for the point you're making, the errors are too negligible for that.

They've done some research on the 5128 already, and produced an extrapolated target that you all at headphones.com have derided. Maybe before you poo-poo it, how about demonstrating that your target is any better with at the very least a lower mean predicted preference rating error than the 5.8 points that this 5128 target from Harman has in relation to the predicted ratings from the original GRAS measurements and Harman target as Sean has shown.

A properly designed target for the 5128 won't have a lower predicted preference rating error than what Harman obtained for a given set of headphones precisely because the method Harman used is intrinsically designed to produce the lowest error possible, on average, for that set. It's a purely logical outcome of the methodology used. Now the issue is that using a different set of headphones, they'd have gotten a different target (particularly when comparing coupling insensitive headphones to coupling sensitive ones), and of course now you'd get a larger error with the first set.

You can visualise this problem fairly easily by dividing the set of 20 headphones Harman used to produce their 5128 target in two subsets of 10 headphones and apply the same methodology for both :
two subsets.jpg

Here very deliberately choosing to put the majority of coupling insensitive headphones in one and coupling sensitive headphones in the other.

Since the sample of 20 headphones Harman used to design this 5128 target comprises a majority of coupling sensitive headphones, you end up with the rather weird situation where their 5128 target, in spite of its low predicted score error vs the predicted score on 711, is not coherent with Harman's own listening tests, for reasons already developed in this thread (using HP.com's 8dB / tilt in that post, things would be a bit different now that they're rather aiming for a 10dB/tilt).

You considered above that HP.com's target is an outlier vs. the other targets put forth, well, Harman's own 5128 target would be just as much.

So yep it should be poo-pooed, it's a particularly nonsensical methodology.

Now the thing is, since HP.com's target also is quite a departure from the results of Harman's listening tests, you could leverage the same criticism against it, at least until additional listening tests have been performed, which is why as far as I'm concerned, I'd be a bit more comfortable with LMG's target for now as a preliminary target, until proven otherwise. You're just late to the party.

Of course even that's a fairly low bar, and what you'd actually need to empirically substantiate the casually thrown around claim that the 5128 provides a more accurate representation of human headphone perception/preference than GRAS is a correlation between predicted and actual preference ratings of greater than the high 86% Harman achieved using their GRAS rig and target for it

I sense a confusion here between assessing whether or not the 5128 is a more accurate representation of how headphones behave on real humans, which should be evaluated with in-situ measurements, and whether or not the combination of a fixture, a target and a predictive model are more effective than others at predicting people's preferences.

I find it quite funny that you always seem to qualify the r number Harman's predictive model resulted in as "high" as if you had to convince yourself that it's a great achievement, I'm still puzzled why some people express it as a percentage, which doesn't make a lot of sense to me, and I am not expecting you to intelligently engage with anything I'd write about it, but if the following was a predictive model designed to rank people by height from tallest on the left hand side to lowest on the right hand side, I think most of us would consider it rather perfectible to say the least.

Screenshot 2023-07-03 at 09.19.31.png
Screenshot 2023-07-03 at 09.21.21.png

That's what the OE model achieved, and that's before any consideration of coupling issues come into play.

When you combine the degree and the distribution of errors the model makes with this (ie inter-individual variation at the blocked ear canal entrance) :
Screenshot 2023-06-24 at 22.44.45.png


You end up with something of very little value in my opinion, even if its not unreasonable to think that this is the best publicly available predictive model we know of today.

Ie, if I show you this :

Screenshot 2021-09-17 at 08.49.09.png
FXK4-tzUUAA05Lp.jpeg


You basically have no way of knowing, using the predicted score, which one of these will actually be preferred by a majority of listeners, as you don't know the error the predictive model made for each (degree and direction, ie positive or negative), and don't know from the score alone their in-situ behaviour.

(as well as the high 85% virtualization correlation,

Harman's validation for the OE virtual methodology relies on this data, with the scores obtained with the real headphones in blue, and the scores obtained with the virtual method in green (with the r number expressed as it should, not in percentages) :

Screenshot 2023-07-03 at 09.59.47.png

The tests actually included a larger number of participants, but only some of them participated in both tests, if we include them all the results are the following :
Screenshot 2023-07-03 at 10.07.00.png


I have not digitised the data for that already but I'm pretty certain that I'd get similar if not worse errors than the predictive model above once scaled similarly (which might not be that easy to do properly...).

This is a fascinating article, but rather for the questions it raises than the conclusions it seems too eager to reach in my opinion.

For example, why is it that in regards to HP4 (K550), the authors write this :
Screenshot 2023-07-03 at 10.04.58.png

When whether using the data from all listeners or only the ones who participated in both tests, it's actually the HPs that scored the most similarly, even though we know from Harman's own research that the K550 is quite susceptible to coupling issues and measures in situ quite differently from how it measures on ear simulators (and therefore how it would have been reproduced by the virtual HP method). But is it really scoring the most similarly or are we seeing the effects of the listeners not using the scale in the same manner between both tests, among other variables ? If so, how can we scale it properly without the presence of anchors in both tests ? In comparison, why is it that no tentative explanation was given for HP5's scores, even though it's the one with the most discrepancy ?

and that was with the old KB007X too, before they moved to more anthropomorphic pinnae which would likely increase this correlation even higher).

That's possible, but also speculative, particularly given the set of headphones Harman used for the virtualisation validation test. For one of them (K550), it's actually one of the few headphones for which the Welti pinna was less representative of the on-head behaviour than the older pinna.

There's a reason why Harman is now very much focused, among other considerations, on the question of coupling issues, they've finally realised that it's a major issue, particularly for the type of headphones they intend to sell. They wouldn't bother to perform in situ measurements if measuring headphones on a fixture and running a predictive model was all you needed to predict preferences and if their virtual HP methodology was all you needed to subjectively evaluate a particular pair of headphones.
 
OP
Robbo99999

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
7,062
Likes
6,934
Location
UK
Straw man. We're not talking about the DFHRTF calculation, we're talking about the target actually used to display and compensate measurements, and the default target Crinacle displays for the B&K on his graph tool is this:
View attachment 296347
i.e. DFHRTF with a -8 dB (-0.8 dB/octave) slope and a +10 dB bass shelf, relative to DF. Here it is not relative to DF (not completely unsmoothed by default either):
View attachment 296348
Compare with your target (dashed line) that has a -10 dB slope and no shelf:
View attachment 296344
Clearly these are considerably different. And if you're weakening your argument to say they have generally similar 'fine-grained' features, 1) that's pretty inconsequential, not only psychoacoustically (high-Q, low amplitude => less audible), but especially in comparison to the large coarse-grained differences and 2) that's not surprising considering Crinacle is sponsored by headphones.com. And yet despite this it looks like even he realizes the target sounds wrong (as many have said, thin/bright/shouty etc.) without a bass shelf (which funnily enough is also a tacit admission that his own Truthear Red he 'tuned' lacks bass, as they fall significantly short of his target down there stock, w/o resistor). Also, here's a bonus game of spot the odd one out:
FqkNSfuaYAA27xz


Another straw man, that's nice of you to give the first one a friend to play with. The pluralist nonsense you're actually planning in order to make headphones/IEMs look better is even worse.

Ironic, considering you're clearly not understanding then rejecting what people have repeatedly been trying to tell you ever since you first got the 5128: raw data is of little use without proper processing, and in this case that means compensation to a target that has direct demonstrated correlation with preference in the form of controlled blind listening tests. This is paramount. Without this, flawed supposedly 'based in theory' target concoctions won't cut it.

They've done some research on the 5128 already, and produced an extrapolated target that you all at headphones.com have derided. Maybe before you poo-poo it, how about demonstrating that your target is any better with at the very least a lower mean predicted preference rating error than the 5.8 points that this 5128 target from Harman has in relation to the predicted ratings from the original GRAS measurements and Harman target as Sean has shown. Of course even that's a fairly low bar, and what you'd actually need to empirically substantiate the casually thrown around claim that the 5128 provides a more accurate representation of human headphone perception/preference than GRAS is a correlation between predicted and actual preference ratings of greater than the high 86% Harman achieved using their GRAS rig and target for it (as well as the high 85% virtualization correlation, and that was with the old KB007X too, before they moved to more anthropomorphic pinnae which would likely increase this correlation even higher).
It's gonna be hard for Resolve & Co to produce a study that has the depth & rigour that the Harman Research achieved, we might have to acknowledge that reality, and therefore accept that they will continue with the project anyway. I'm sure they will attempt to make it as rigorous as possible within the confines of what a group of knowledgeable headphone enthusiasts (& reviewers)/& technical experts like @Mad_Economist can achieve whilst being backed up by their headphone review company. Who knows, maybe it'll end up being highly rigorous with much depth, but we will have to wait & see. Either way I'll be trying out their finished product on my headphones, and I'm looking forward to trying it out - I'd have to be wary of the "honeymoon period" when trying it out, as I always have a tendency for early enthusiasm when trying a new EQ or new headphone!
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,833
Likes
243,185
Location
Seattle Area
The previous measurement paradigm essentially ignored HpTF effects - at least when it came to the way people commonly evaluated them - and I have yet to come across a compelling reason why we as a community shouldn't want that data. So far what I'm seeing is some version of "I don't understand it, so I don't want it".
We don't want two versions of the truth. It creates confusion and no one then knows what is what resulting in people ignoring both sets of data. We create standards so that everyone unifies around one. Having two competing standards means no standard.

And what do you mean "I don't understand it?" I had the 5128 for a good while. I was ready to buy it but its issues were immediate and present. I let Mad_Economist create a target for me using DF from B&K and it was a bust. Screwing around with adjusting it this and that way without large scale, controlled testing is just random playing. There is no logic around wanting more accuracy above 8 to 10 kHz when so many other variables conspire to screw things up.

I realize you want to salvage the investment but best path forward would be to sell the darn thing and help get the entire audio industry rallying around one standard. The one that is heavily researched across many years.
 

crinacle

Member
Reviewer
Joined
Jul 31, 2019
Messages
60
Likes
795
Location
SG
that's not surprising considering Crinacle is sponsored by headphones.com. And yet despite this it looks like even he realizes the target sounds wrong (as many have said, thin/bright/shouty etc.) without a bass shelf (which funnily enough is also a tacit admission that his own Truthear Red he 'tuned' lacks bass, as they fall significantly short of his target down there stock, w/o resistor).

The stock RED tuning has always been my ideal. The Bass+ adapter was included just in case consumers did indeed prefer a higher bass shelf (because there were so many armchair objectivists complaining that it was going to be too little based on early measurements), but at the end of the day it seems that the vast majority of users prefer it stock.

1688393590645.png

Just as I had tuned it from the very beginning.
 

Mad_Economist

Addicted to Fun and Learning
Audio Company
Joined
Nov 29, 2017
Messages
578
Likes
1,706
Spirited discussion here it seems - it's a long holiday weekend in the land of freedom, so you'll pardon me if I refrain from working too hard in my replies.

Another straw man, that's nice of you to give the first one a friend to play with. The pluralist nonsense you're actually planning in order to make headphones/IEMs look better is even worse.
Re "pluralist nonsense" - glad we waited out Pride for that framing - we're referring consistently to Segmentation of Listeners Based on Their Preferred Headphone Sound Quality Profiles By Olive, Welti, and Khonsaripour here. In general, we are specifically talking about enacting what's highlighted here from Sean's conclusion
1688398590402.png

In general, the goal here is not to "make headphones/IEMs look better", but to come up with something that better matches the subjective impressions of certain subgroups of listeners, based on the data that Sean has kindly provided. None of this is meant to indict the Harman target, and indeed the predictive power of the Harman target and Sean's statistical models is and will remain a canonical internal reference point for us - we are not looking to publish something that makes things muddier, except in the areas where people are overfitting.

A major fundamental issue, mentioned by Resolve himself in Twitter, as well as Crinacle in his video - to the point where he advises against direct treble comparisons of BK5128 measurements, is the difficulty to achieve consistent insertion on BK5128. On IEC711 this is trivial. I haven't heard anyone with the 5128 at hand finding a solution to this issue.

Re: insertion depth: it must be noted that the above-bolded statement is incorrect as a generalization. It is easier to achieve consistent insertion depth with a straight or conical metal canal extension, but the first 10mm of ex. a KB5010 canal are geometrically very similar to a Type 4620 canal (because both are based on averages of human ear geometry), and there's no reason that the 60318-4/711 coupler will make getting a consistent depth easier than the 4620 there.

Comparing 60318-4s with metal canal extensions to the Type 4620 in the 5128, it's true that it's harder to force the same insertion depth with the 5128. This is because different IEMs fit that canal differently, much as they would with real listeners. Some people have raised that this would make it harder to observe the Q/damping thereof of the length modes of the IEM with the 4620, but this same argument can be used to argue for a 2cc or .4cc coupler over a 60318-4 - and I don't think that's disqualifying. Perhaps "damping of length mode Q" is a parameter that merits specific attention, since there's certainly meaningful variation among IEMs there. But you do not necessarily need a 60318-4 for it, and indeed you arguably would benefit from using a different coupler instead.

And what do you mean "I don't understand it?" I had the 5128 for a good while. I was ready to buy it but its issues were immediate and present. I let Mad_Economist create a target for me using DF from B&K and it was a bust.
I'm a bit curious about what you'd have preferred on that front - the Harman average delta approach would have left the same issues raised in the thread (inaccurate low frequencies due to leaky coupling, positionally varying high-frequency peaks), which also exist with the 45CA (albeit it's easier to get a seal with).

They've done some research on the 5128 already, and produced an extrapolated target that you all at headphones.com have derided. Maybe before you poo-poo it, how about demonstrating that your target is any better with at the very least a lower mean predicted preference rating error than the 5.8 points that this 5128 target from Harman has in relation to the predicted ratings from the original GRAS measurements and Harman target as Sean has shown

Re: testing other targets in (variants of) the Harman predictive model: I think that's a great idea, and we're working on that presently. We've had some internal methodological disagreements (e.g. should the response measurements from the 5128 be "the response of the headphones in question themselves", "the response of the used replicator headphone from Olive's work equalized to match the headphones on the 5128", or "the response of the replicator headphone equalized to match the headphones on the 45CA/43AG, measured on the 5128", but ultimately the latter has won), and there's some interest in doing separate listening tests to avoid potentially overfitting to the dataset we have presently, but none of that IMO should stand in the way of this sort of comparison. Expect it in the fairly near future, although it depends on how much stuff ends up on my plate!
 

markanini

Major Contributor
Joined
Feb 15, 2019
Messages
1,816
Likes
1,892
Location
Scania
The stock RED tuning has always been my ideal. The Bass+ adapter was included just in case consumers did indeed prefer a higher bass shelf (because there were so many armchair objectivists complaining that it was going to be too little based on early measurements), but at the end of the day it seems that the vast majority of users prefer it stock.

View attachment 296490
Just as I had tuned it from the very beginning.
I certainly expected to prefer it with the bass adapter, based on prior sets I liked. However after listening to my playlist of familiar songs, the leaner bass sounded cleaner, and I can hear texture that I generally miss in IEMs vs headphones and speakers.

The curious catch is that when I fit the included wide bore tips, instead of the narrow bores that I prefer, my preference changes and I prefer the bass adapter.
Maybe it's no mere coincidence that for Truthear Hola, which has a response very closely resembling the TE Red with the bass adapter, I preferred the narrow bore tips included too.

I think it's important to understand that while the development and existence of established targets is very valuable, Headphones and IEMs in sense are only a rough approximation of a stereo speaker setup. That warrants diverging from strict granular target adherence. Some slight differences in HF energy might change preferences, and this might be obscured to measurement systems.

If I may be so blunt I think you found that a lot of people like a 2dB less U-shaped response vs. Harman 2019, when the treble is clean and smooth.
 

markanini

Major Contributor
Joined
Feb 15, 2019
Messages
1,816
Likes
1,892
Location
Scania
@Mad_Economist
I'm saying something directionally correct. Compared to BK5128 IEC711 systems have more freedom to choose, and reproduce, insertion depth. This is what operators of each system have commented, multiple times.

Same old song and dance from you where you take comments apart, word by word, to object, only after changing to context to to something irrelevant to the subject matter, or the actual statement.

As far as headphones Sean Olive addressed the issue with BK5128 not offering an improvement in representing human subjects :
Untitled.png
 
Last edited:

Mad_Economist

Addicted to Fun and Learning
Audio Company
Joined
Nov 29, 2017
Messages
578
Likes
1,706
@Mad_Economist
I'm saying something directionally correct. Compared to BK5128 IEC711 systems have more freedom to choose, and reproduce, insertion depth. This is what operators of each system have commented, multiple times.

Same old song and dance from you where you take comments apart, word by word, to object, only after changing to context to to something irrelevant to the subject matter, or the actual statement.
Except that it...isn't, as a generalization, which is what I was responding to. It's true that some 60318-4/711 configurations give greater precision in insertion depth, but that's not the case with the anthropometric ear canals, and indeed more true-to-life fitting was part of GRAS' sales pitch for them.

Something I was attempting to be vaguely oblique about but will now state directly is that being unable to force unrealistic insertion depths is not a bug, it's a feature. It's true that it would be preferable to have a variety of canals (different type 4620 pinnae when, BK?), but if you have one, having it realistically interact with the geometry of the actual device is preferable. If you want to look at the length mode damping behavior alone, you're better off omitting the parallel resonators and using a .4cc.
 
Status
Not open for further replies.
Top Bottom