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

[thewas]

To be clear, I'm not quite advocating for a 2 shelf like Harman, but I do advocate for a variable slope

Thank you for your reply and I agree by finding the 2 shelving filters of Harman a too strong limitation.

 

[Dazerdoreal]

Recently, Crinacle posted the measurements of his IEMs on the 5128. (https://crinacle.com/graphs/iems/graphtool/?share=Diffuse_Field_(-0.8dB_tilt)_Target,Dioko) Out of curiosity, I did the following test:

1. I used AutoEQ to adjust the 5128 measurement to the "Diffuse Field -1.0dB tilt" target.
2. I saved the EQ preferences.
3. I applied the saved EQ preferences to the "old" measurement graphs using the IEC 60138-4 standard.

I have attached some of the results.

From my observations, it seems that the 5128 and IEC 10138-4 measurements are not that different from each other in most areas. However, what varies significantly is the target that they use. The tilted Diffuse Field targets emphasize less bass, more lower mids around 300hz, a peak at 2.5 kHz, and a subsequent dip on a Harman scale. This signature would basically remain consistent even if a different "tilt" were chosen.

When I first heard about the upcoming 5128 target on YouTube videos, I thought that it would make the Harman curve obsolete, as the measurement methods were fundamentally different. However, after looking at the initial results, it seems that it is still possible to create a target that is close to a Harman curve-ish signature, if one desired to do so. (Maybe the differences on over-ears are bigger though.)

 

[Robbo99999]

Nowadays very rarely listening with headphones, but out of curiosity tried his HD 600 EQ few days ago but didn't like it as much as for example Oratory's ones, don't think though that the reason for this is his 5128 but rather current constant slope target (which he is considering of changing to 2 shelving filters like Harman).

What about you, how is your preference with your headphones? (I apologise in advance if you posted it already in this thread and I missed it).

Interesting to get your feedback, thanks for that. My preference with the Resolve EQ on my HD560s was to just put in a tiny +1.5dB Low Shelf at 70Hz (Q0.707). This allowed me to follow the bass line in the music and also balanced it better with the rest of the frequency response. I tried using a linear slope to get the same effect, but I didn't want to change the treble, just bumping up the bass was the best solution for me. I also decided at 70Hz rather than 105Hz because I didn't want to do anything to change the character of voices, which I thought was spot on.
(Paradoxically (although praps not particularly paradoxically!) I enjoy Harman EQ's too, but comparing Resolve EQ with Harman EQ on my HD560s I preferred that slightly bass modified version of Resolve's EQ.)

To be clear, I'm not quite advocating for a 2 shelf like Harman, but I do advocate for a variable slope - or at least a boundary window around the slope at key regions. This is sort of what I was getting at in the interview we did with Dr. Olive, where I think a 3 point preference adjustment would be good, one specifically at the highest point of the ear gain around 3khz. This is in part due to the additional adjustment around there in the later research, which he confirmed was based on listener feedback as well. He seemed to think it was a good idea, but also made a good point that if you adjust bass and treble you effectively also adjust the midrange. So maybe my issue could alternatively be solved be shifting the treble adjustment up slightly.

Regarding shifting the treble up, I certainly wouldn't want that on the HD560s EQ - I mean I know it's only one data point (me), but just wanted to say.

 

[_thelaughingman]

Utter curiosity of mine. Would adding a 1 db bass Tilt to the Diffuse field curve help the target achieve similar response as the Harman target? Is it already factored in to the target?

 

[Resolve]

Recently, Crinacle posted the measurements of his IEMs on the 5128. (https://crinacle.com/graphs/iems/graphtool/?share=Diffuse_Field_(-0.8dB_tilt)_Target,Dioko) Out of curiosity, I did the following test:

1. I used AutoEQ to adjust the 5128 measurement to the "Diffuse Field -1.0dB tilt" target.
2. I saved the EQ preferences.
3. I applied the saved EQ preferences to the "old" measurement graphs using the IEC 60138-4 standard.

I have attached some of the results.

From my observations, it seems that the 5128 and IEC 10138-4 measurements are not that different from each other in most areas. However, what varies significantly is the target that they use. The tilted Diffuse Field targets emphasize less bass, more lower mids around 300hz, a peak at 2.5 kHz, and a subsequent dip on a Harman scale. This signature would basically remain consistent even if a different "tilt" were chosen.

When I first heard about the upcoming 5128 target on YouTube videos, I thought that it would make the Harman curve obsolete, as the measurement methods were fundamentally different. However, after looking at the initial results, it seems that it is still possible to create a target that is close to a Harman curve-ish signature, if one desired to do so. (Maybe the differences on over-ears are bigger though.)

Absolutely. But there's a difference between doing that for yourself and identifying that as a 'reference curve'. As we've discussed in a few places, it's not possible to simply transpose Harman to the new system by using a group of headphones, or by using EQ profiles on a particular headphone (with one exception, but that's not likely to happen), because the headphone-to-ear transfer function isn't going to be the same.

 

[Robbo99999]

Utter curiosity of mine. Would adding a 1 db bass Tilt to the Diffuse field curve help the target achieve similar response as the Harman target? Is it already factored in to the target?View attachment 276180

Resolve has already applied a Linear 8dB tilt (from 20Hz to 20kHz) to the Diffuse Field Curve of the 5128 which is supposed to mimic the overall "change slope" that is used by the Harman Curve. I'm just mentioning that because it sounds like you've misunderstood that, or didn't know that it had already been applied to the Diffuse Field Curve. I also don't think the goal of the B&K Target is to be as close as possible to the Harman Curve, I don't think that's one of the goals, even though I think it makes sense that he applied the Linear 8dB tilt (similar slope to Harman) to the Diffuse Field Curve.

 

[Robbo99999]

I haven't really been using my Anechoic Flat EQ'd & Room EQ'd JB308p Mkii speakers (& sub) much for listening to music over the past weeks, but I did yesterday night - and now I'm wondering if the Resolve B&K EQ is too thin & bright - I think I'm getting more bass and less treble from my "reference speaker & sub" system. My previous reports of finding the Resolve B&K EQ better than a Harman EQ on my HD560s were definitely genuine when I reported those experiences over the last week, but when talking about accuracy to a reference speaker system I'm less certain that it's close. I haven't done enough testing to discern which I think is most accurate vs my speaker system - the Resolve B&K or the Harman 2018 Curve.......but thinking about my experiences last night I'm thinking the headphones need less treble in both whilst having more bass than the Resolve B&K - this leads me to surmise that the 2013 Harman Curve could be the most accurate because it has less treble than Harman 2018 and more bass than Resolve B&K.....my next port of call could be testing a 2013 Harman EQ vs my speakers and Resolve B&K. Don't know yet when I'm gonna do this though. Headphones are so darn difficult to be objective on!

EDIT: Actually my Harman EQ for my HD560s sound more accurate to my speaker system (mainly assessed through bass and vocal character on my known reference tracks) than the Resolve B&K EQ. I should say that the Harman EQ for my HD560s has always been a slightly modified version, either adding 1dB more bass (amoungst a few other smaller changes) on the Oratory EQ or another Harman EQ I use which is essentially a linear 1.25dB downwards tilt on an Oratory EQ. I do like the Resolve B&K EQ though and I think it captured me through a slightly different presentation to that which I was used to, probably in treble detail with less emphasis on bass.......but I think a slightly modified Harman EQ is more accurate to my speaker system. I'm not gonna bother testing out a Harman 2013 EQ at the moment though as I'll need to dig out REW and do some mods. My original feedback on the Resolve B&K EQ re my HD560s still stands though inasmuch that I always still agree it needed a bit more bass.....but I think I've reverted to using my long time Harman EQ's for my HD560s (which in the past I had historically compared against my speakers, just I hadn't revisited that comparison until recently). I should note that I should try the Resolve B&K EQ's out on my HD600 though, yet to try it on those.

EDIT#2: you can probably see that there is a real spirit of "circle of confusion" that I've exemplified in my posts, re changing my mind & wondering which is most accurate, etc.....it's not easy! I generally always eventually gravitate to an EQ after a few weeks of testing though, as I originally noted somewhere in my posts either on here or on Resolve's website, so honeymoon periods are totally real! (first post in thread updated just to note this change).

EDIT#3: I think I'm gonna have to retract some of my previous statements in this thread where I've theorised or stated that the B&K Target is more accurate than the Harman Target. I mean I've still got to test out my HD600, but that's where I'm at for my HD560s.

 

[oratory1990]

Forgive me but if some of them were 'very old' what purpose does this graph serve, given the HD650 has gone through various revisions over the years? Myself and many others have demonstrated simply the different screen used (black paper, silk, or current silver screen) changes the frequency response and the driver itself has probably changed over the last 20 years.

I was unable to show any correlation between age, mesh color and measurement results.
The only correlation was in the age of the earpads (duration of usage, not year of manufacturing).

 

[Resolve]

EDIT#2: you can probably see that there is a real spirit of "circle of confusion" that I've exemplified in my posts, re changing my mind & wondering which is most accurate, etc.....it's not easy! I generally always eventually gravitate to an EQ after a few weeks of testing though, as I originally noted somewhere in my posts either on here or on Resolve's website, so honeymoon periods are totally real! (first post in thread updated just to note this change).

It may change again!

One thing to keep in mind, as Oratory has pointed out, even if you zero out the pre-gain, there's still an SPL difference between EQ on and EQ off - and also Harman (I ran into this issue as well when trying to compare with my Harman profiles). So it's best to spend more time with a profile and a range of specific tracks, as opposed to toggling among them and assessing that way.

 

[Robbo99999]

It may change again!

One thing to keep in mind, as Oratory has pointed out, even if you zero out the pre-gain, there's still an SPL difference between EQ on and EQ off - and also Harman (I ran into this issue as well when trying to compare with my Harman profiles). So it's best to spend more time with a profile and a range of specific tracks, as opposed to toggling among them and assessing that way.

Ha, yes, possibly! (re changing again!) Negative Preamp was the same for all the EQ's I was comparing but I couldn't say they were level matched, subjectively they were the same volume. I mean I did spend a lot of time with the B&K EQ, I was listening to it exclusively after my initial comparison for a few days over a few sessions - just when I listened to my "reference speakers" last night I noticed a difference and then immediately went back to revisit the B&K EQ, and then subsequently the slightly modified Harman EQ.....then I realised the modified Harman EQ was closer to my speakers. I know myself well enough that I can flip flop between EQ preferences in the short term, but I tend to stabilise on one after some initial flip flopping! I do think the work you're doing is a worthwhile endeavour though, anything where anyone is trying to push forward headphone fidelity cannot be frowned upon, and it's certainly interesting for folks to check out and compare. Did you ever think about wacking the B&K in a really well setup listening room with some totally optimised speakers and then measuring the response at the eardrum to calculate a starting Target Curve? I think Harman rotated the head from central and then to +/-30 degrees to then average the curves to create their starting point. (That's a different approach to your Diffuse Field Curve + linear harman style room slope). The advantage of that would be that it's the same approach as Harman but instead with an arguably more anatomically accurate mannequin - perhaps the end result would be that it would sound similar to Harman but with greater "depth & nuance / authenticity" due to slightly more accurate parts in some areas of the frequency response.

EDIT: certainly if I flip back to preferring the B&K Target then I will of course say so here in this thread......and I've still got the HD600 to test as well to see what conclusions I come to re that particular headphone and B&K Target/Harman Target/etc.

 

[Resolve]

The advantage of that would be that it's the same approach as Harman but instead with an arguably more anatomically accurate mannequin - perhaps the end result would be that it would sound similar to Harman but with greater "depth & nuance / authenticity" due to slightly more accurate parts in some areas of the frequency response.

Yeah, we covered this in an earlier post on the topic. For sure we could do this (as could anyone with a 5128), and I think if Harman were to do it for a new in-room response with the 5128 and then apply the filters that would be a better approach than their internal stop gap of using 20 headphones on both systems.

But the benefit of doing it our way is that it's free from any additional baggage that would come as a result of using a particular pair of speakers and a particular room. So it's both more compatible with the wider body of science out there (and anchored to DF), and also free from any noise that may be carried forward as a result of specific test conditions. Like, the Harman approach is totally valid in my opinion, but even calculating in-room bakes the Harman speakers and room conditions into the research, and that gets carried forward into the rest of the results and any further research that gets done on this topic.

As Dr. Olive has pointed out, you could do a lot worse than their setup, but this just gets down to a point of methodological difference where we'd rather remain free from any conditions, even if those speakers are pretty good - merely for the sake of not baking them into the target moving forward.

Edit: Point being... while we could do that, and it would be interesting, we'd also have that same issue.

 

[Robbo99999]

Yeah, we covered this in an earlier post on the topic. For sure we could do this (as could anyone with a 5128), and I think if Harman were to do it for a new in-room response with the 5128 and then apply the filters that would be a better approach than their internal stop gap of using 20 headphones on both systems.

But the benefit of doing it our way is that it's free from any additional baggage that would come as a result of using a particular pair of speakers and a particular room. So it's both more compatible with the wider body of science out there (and anchored to DF), and also free from any noise that may be carried forward as a result of specific test conditions. Like, the Harman approach is totally valid in my opinion, but even calculating in-room bakes the Harman speakers and room conditions into the research, and that gets carried forward into the rest of the results and any further research that gets done on this topic.

As Dr. Olive has pointed out, you could do a lot worse than their setup, but this just gets down to a point of methodological difference where we'd rather remain free from any conditions, even if those speakers are pretty good - merely for the sake of not baking them into the target moving forward.

I totally get that if Harman were to measure the B&K 5128 in their listening room and then use that as their fundamental starting point then that listening room gets baked into the equation so-to-speak, but you could use your own choice of listening room (or even hire a perfect studio?)...........but I understand your point when you stress remaining free of ANY conditions (rooms), ie choosing Diffuse (saying there is no room, EDIT: and no point source), but that's also setting you with a starting point stipulation that could be an arguably more non-optimal starting point - and afterall you apply a generic room after that initial measurement by applying the 8dB linear slope anyway which could defeat the purpose somewhat. It's the sort of thing I'd experiment with anyway, I'd probably try both options and see what comes out of it. I actually think it makes a lot of sense to bake in a great room.

EDIT: you could do a number of experiments where you bake in a number of great rooms to see if it makes any appreciable difference to the results/listening experience - so basically wacking the B&K 5128 into a handful of fantastic studios or otherwise unanimously well measuring rooms. Maybe something could be learned by that in terms of the influence or importance of the room as it passes through the "simulation" of the dummy head & speaker to headphone approximation. Or take an average of those great rooms whilst kicking out the outliers.

 

[solderdude]

not year of manufacturing

2018 might have been an exception if we were to believe this (and I do):

Solving the HD650 mystery

I was initially planning to post this on our company blog, but it took too much time to get comments from Sennheiser, so I shelved the article. Some...

www.superbestaudiofriends.org

I too found a deviation in just 3 HD650 over the years (all fitted with new pads)

early black screen (mine), silver screen (2017) and 2014 silverscreen HD650 which sounded 'veiled'.
Strangely enough the 2014 'veiled' one acted a lot like the 2018Q4 in the article above.... assuming the owner did give me the correct manufacturing date.

Of course large age difference but all with new pads. Quite some production spread. In the same vein as what you found I reckon. All measured on the same 'fixture'.
Not all HD650 are created equal but this has been known for a while... still to keep production that similar over so many years is still quite an achievement. Sennheiser probably also had to deal with suppliers changing or stopping production of materials over that time period.

 

[amirm]

But the benefit of doing it our way is that it's free from any additional baggage that would come as a result of using a particular pair of speakers and a particular room. So it's both more compatible with the wider body of science out there (and anchored to DF), and also free from any noise that may be carried forward as a result of specific test conditions.

What body of science? There is none to correlate diffused field response of speakers to listener preference. Indeed, power response that is derived for that shows negative correlation with listening tests. A few papers using diffused field for headphone research doesn't remotely come to same level of research we have from Harman. Diffused field testing of HATS was started when we knew much less about listener preference than we do today. Using it as an element for anything is just wrong. And certainly doesn't point to a "wider body of science." Nothing is remotely as researched as Harman's work across a dozen papers and countless headphones and listeners. To use that, you need to use their GRAS fixture. Going anywhere else means you are on your own with no body science defending you other than a couple of papers...

 

[amirm]

EDIT: you could do a number of experiments where you bake in a number of great rooms to see if it makes any appreciable difference to the results/listening experience - so basically wacking the B&K 5128 into a handful of fantastic studios or otherwise unanimously well measuring rooms. Maybe something could be learned by that in terms of the influence or importance of the room as it passes through the "simulation" of the dummy head & speaker to headphone approximation. Or take an average of those great rooms whilst kicking out the outliers.

How do you determine those are "great rooms?" Harman used an IEC compliant room so results can be replicated and research confirmed. Taking random rooms and claiming to be great itself is subjective and cannot be the basis for any standard.

 

[Resolve]

EDIT: you could do a number of experiments where you bake in a number of great rooms to see if it makes any appreciable difference to the results/listening experience - so basically wacking the B&K 5128 into a handful of fantastic studios or otherwise unanimously well measuring rooms. Maybe something could be learned by that in terms of the influence or importance of the room as it passes through the "simulation" of the dummy head & speaker to headphone approximation. Or take an average of those great rooms whilst kicking out the outliers.

This would be interesting to do, but not necessarily for the purposes of developing a reference curve. Again I think it comes down a difference of approach, and for a practical test application I don't really have any issues with the way Harman did it. It's more a question of how much you value wider compatibility. I think of it kind of like... our approach is a bit more condition-agnostic or 'de-siloed', with the ability to scale the concept to far more test situations, being backwards compatible with flat DF and scalable to future equipment as well, while Harman is specific to that test environment and equipment and will always carry those variables forward until they do the research again.

The other thing is that, in the event we do practical controlled tests with this and end up with a variable slope or preference boundaries like I'm advocating for, it can still be backwards compatible.

If you look at the other research I referenced, you can see that while most of the slopes are variable to some degree, the Harman slope is the outlier being the most variable. We're not saying that's wrong... rather that it's worth considering what this looks like without those conditions. The one thing that seems consistent is the bass to treble delta that people prefer, and so it makes sense to go with that as the starting point.

Edit: And... just so folks properly understand this, DF and Harman in-room before the filters were applied is very similar (if the same smoothing gets applied). There was that additional preference adjustment to the top of the ear gain, and this is precisely where I'm going with preference boundaries around the target as well - the whole point of my thread on this topic.

 

[Robbo99999]

How do you determine those are "great rooms?" Harman used an IEC compliant room so results can be replicated and research confirmed. Taking random rooms and claiming to be great itself is subjective and cannot be the basis for any standard.

What's an IEC compliant room? That is certainly a gap in my knowledge. I was simply pointing out that it might not be the best starting point to ignore the influence of the room on the initial target measurement. As I understand speaker & room interaction we know we like anechoic flat speakers in a room that has good effective room EQ below the transition zone, so I was really pointing towards those variables being satisfied.....but I suppose there's additional considerations to be taken into account beyond that, which might be associated with your IEC compliant room?

 

[Robbo99999]

This would be interesting to do, but not necessarily for the purposes of developing a reference curve. Again I think it comes down a difference of approach, and for a practical test application I don't really have any issues with the way Harman did it. It's more a question of how much you value wider compatibility. I think of it kind of like... our approach is a bit more condition-agnostic or 'de-siloed', with the ability to scale the concept to far more test situations, being backwards compatible with flat DF and scalable to future equipment as well, while Harman is specific to that test environment and equipment and will always carry those variables forward until they do the research again.

The other thing is that, in the event we do practical controlled tests with this and end up with a variable slope or preference boundaries like I'm advocating for, it can still be backwards compatible.

If you look at the other research I referenced, you can see that while most of the slopes are variable to some degree, the Harman slope is the outlier being the most variable. We're not saying that's wrong... rather that it's worth considering what this looks like without those conditions. The one thing that seems consistent is the bass to treble delta that people prefer, and so it makes sense to go with that as the starting point.

Edit: And... just so folks properly understand this, DF and Harman in-room before the filters were applied is very similar (if the same smoothing gets applied). There was that additional preference adjustment to the top of the ear gain, and this is precisely where I'm going with preference boundaries around the target as well - the whole point of my thread on this topic.

I think I understand what you're getting at in terms of "backwards compatible & future scaleable", you're saying that if you refine the slope of DF through your experimentation & user feedback then this same slope can be applied to the DF of other measurement fixtures? The reason being because the DF determination of every measurement fixture out there is to a set standard, therefore any modification you make to optimise DF on your 5128 will be applicable to DF on any other fixture? (as long as you ignore the relevance of how close the overall HRTF (head/pinnae/inner ear/eardrum/impedence/etc) is of the various fixtures to a representative human - in terms of how relevant it is when a human wears the headphone that has been calibrated in such a manner - which is important when it comes to measuring the headphone additionally - ie that approach wouldn't work for the miniDSP EARS rig for example due to lack of anatomically accurate or no head & combined with anatomically inaccurate overall ear impedance) I guess that doesn't completely refute the idea I had that this might not be the best starting point, and that an ideal in-room measurement could be the best starting point - whilst also taking into account the IEC compliant rooms that Amir corrected me on (although at this point I don't understand how fundamentally important "IEC compliant" really is - until Amir might elucidate the point).

EDIT: I think I'm right on some these ideas, but I'm not 100%, so I don't mind being corrected on them.

EDIT: probably the biggest spanner in the works is the headphone transfer function variable in terms of you supposing that DF adjustable target is applicable to all measurement fixtures.

 

[Resolve]

I think I understand what you're getting at in terms of "backwards compatible & future scaleable", you're saying that if you refine the slope of DF through your experimentation & user feedback then this same slope can be applied to the DF of other measurement fixtures? The reason being because the DF determination of every measurement fixture out there is to a set standard, therefore any modification you make to optimise DF on your 5128 will be applicable to DF on any other fixture? I guess that doesn't completely refute the idea I had that this might not be the best starting point, and that an ideal in-room measurement could be the best starting point - whilst also taking into account the IEC compliant rooms that Amir corrected me on (although at this point I don't understand how fundamentally important "IEC compliant" really is - until Amir might elucidate the point).

Not exactly... and to be clear, I'm personally advocating that we test a variable slope with the 5128 to see where preference boundaries should fall, because I think it's important to reflect that. But that's not really where the compatibility and scalability benefits come into it necessarily. Merely that if you wanted to use DF + slope on another fixture you could, and there would be a good theoretical justification for doing so - plus being able to go back to straight DF for those wanting that, or apply a different slope... there's a common rig-specific baseline to go from. But of course, we're still having internal discussions on how best to move forward with this 5128 target.

 

[_thelaughingman]

What's an IEC compliant room? That is certainly a gap in my knowledge. I was simply pointing out that it might not be the best starting point to ignore the influence of the room on the initial target measurement. As I understand speaker & room interaction we know we like anechoic flat speakers in a room that has good effective room EQ below the transition zone, so I was really pointing towards those variables being satisfied.....but I suppose there's additional considerations to be taken into account beyond that, which might be associated with your IEC compliant room?

If i’m not mistaken, IEC compliance of a room has quite a few factors associated to construction of the room, allowable pressure and other factors. Anechoic chambers or any room that would conduct testing of audio products in my opinion must go through IEC compliance process to be acceptable.