ZMF Caldera Headphone Review

[solrage]

I remember being bothered by what looked like a pattern in Caldera's frequency dips in lower treble. So did a bit of analysis and this came out:
View attachment 333795
Notice how there is a clear pattern to the peaks and valleys. The peaks are about 2.5 kHz apart. And dips, about 1.4 kHz. The two coincide at 7.5 khz or so resulting in that peak being depressed. Wonder what the dimensions of the drivers are and whether these are longitudinal modes of the driver, or modal response of the cavity. Hard to imagine this being a planned response given its correlation.

Would have loved to measure the driver on Klippel to see if this is an issue of that component. Planar drivers in line array speakers frequently suffer from such modes and requires care to avoid them (see the stiffening knurling that Dan Clark deploys).

Assuming they're modal responses of the driver or cavity (which I do think is possible), I'd question how audible they are. I remember reading in my research on psychoacoustics how the brain tends to filter heavily for room modes at higher frequencies. This is why I'm skeptical of merely EQing a measurement to comply to Harman as it assumes everything that's measured is heard, or that it's heard at the same level/severity of how it's measured. Room EQ programs don't do this. Their EQing is almost always subtler than what you'd think merely by looking at the measurements. I'd be surprised if similar principles aren't at work with headphones.

 

[IAtaman]

Assuming they're modal responses of the driver or cavity (which I do think is possible), I'd question how audible they are. I remember reading in my research on psychoacoustics how the brain tends to filter heavily for room modes at higher frequencies. This is why I'm skeptical of merely EQing a measurement to comply to Harman as it assumes everything that's measured is heard, or that it's heard at the same level/severity of how it's measured. Room EQ programs don't do this. Their EQing is almost always subtler than what you'd think merely by looking at the measurements. I'd be surprised if similar principles aren't at work with headphones.

Maybe the 7K and 10K ones, but the one on 4K, that looks like a 6dB dip with maybe a Q of 2 - I'd be very surprised if it is not audible. And if it is not audible, then why not iron it out in the first place anyway and save everyone 20 pages of discussions whether Harmon target is scientific enough for the people who like Caldera.

 

[L0rdGwyn]

Scientific research is conducted to a very large extent by companies with financial interests in some type of market. I am not referring to HiFi specifically. Research is conducted by pharmaceutical companies, in computer technology, transport, etc. - by capitalist companies with a market interest if you so wish to call it. But one cannot reject scientific results on the basis of the underlying interest. Sure, it controls the priorities, but if you now know something about science as you claim, then you also know that your objection is irrelevant. Then you also know that the criteria for evaluating scientific results rest on correctly applied methods and verifiability.

When it comes to HiFi, it is rather typical that manufacturers do not conduct very much scientific research. Rather, the typical thing is that they want to avoid correct tests and do not present measured results. Rather, they rely on undefined subjective concepts, i.e. in other words they sell flowers. So, if Harman's research has a commercial interest at bottom, it is not unique to commercial companies, but it is quite unique to the HiFi industry.

When it comes to HiFi, as far as science is concerned, I would like to say that there is not very much scientific research going on at all. HiFi rests mostly on scientific research that was carried out a long time ago. I think it's more accurate to say it's about engineering. There are some exceptions, and the dividing line is not crystal clear, but in general it is a matter of technical design rather than proper science. I might even think ASR ought to be called audioengineeringreview (AIR). You see all the time in all the slightly more infected discussions in this forum statements about what science is really about. You can of course discuss that, but fundamentally ASR is about audioengineering - which to a large extent is noting but applied scientific knowledge. And sometimes such established knowledge that you don't even think that once upon a time it was science that created this knowledge.

I don't disagree with you, and I do believe in the scientific method and verifiability, which is why I suggested the Harman research should be verified before being put into practice or measured against as an industry standard.

Here's an analogy that maybe illustrates better my feeling - say Ford over the past five years did a significant amount of research on buyer preference of car color. What they found was 64% of car buyers preferred their car to be black. Then a popular car reviewing publication began dolling out bad reviews if the car was in any color but black given the new research. Furthermore, they began advocating for the entire industry to only release cars in black, because a majority of people prefer it. Companies could also release cars in other slightly different shades of black, but ultimately, a variation of black is the only real option for a positive review. Given all of the bad reviews, car companies actually started doing it, given it was hurting their bottom line.

Now a high-end boutique car company comes along and releases a car in purple. Its niche audience really loves the new color, even though it isn't verified by any research. Regardless, the car publication gives it a bad review, predictably, but says if you spray paint it black, then it gets a full recommendation! They also go onto say, if you follow our testing methods, you will find that your preference actually IS black, you just aren't aware of it.

Do you see why that might be irritating to the niche audience of the purple car company?

Continuing on with my car analogy (I'm not even into cars, but a lot of people seem to be), using tubes in your audio gear is like driving a classic sports car. It isn't the best performing, it doesn't even have power steering, it gets horrible MPGs and overall is terrible inefficient. But it looks cool as hell and is still extremely fun to drive, compared to a sports car with contemporary engineering and technology.

 

[solrage]

Maybe the 7K and 10K ones, but the one on 4K, that looks like a 6dB dip with maybe a Q of 2 - I'd be very surprised if it is not audible. And if it is not audible, then why not iron it out in the first place anyway and save everyone 20 pages of discussions whether Harmon target is scientific enough for the people who like Caldera.

Why would you be surprised if it's not audible? I mean, I posted another measurement in which the dip doesn't even exist and measures the same in that region as the Stealth. I think there are two issues in terms of if it's audible and, if so, how much. I do think it's a mistake to assume it's as audible as the measurement suggests. As for "why not iron it out in the first place," that assumes that, one, it's easy to do or that, two, doing so won't introduce other (perhaps more audible) issues.

I do plan on actually testing this myself the next time I use the Caldera. I'll be able to blindly switch a 4k EQ filter on and off. My prediction (which I'm open to being completely wrong about) is that if the 4k dip is audible it won't be nearly as severe as the measurement and overcorrecting for it will sound worse.

 

[solrage]

I don't disagree with you, and I do believe in the scientific method and verifiability, which is why I suggested the Harman research should be verified before being put into practice or measured against as an industry standard.

Here's an analogy that maybe illustrates better my feeling - say Ford over the past five years did a significant amount of research on buyer preference of car color. What they found was 64% of car buyers preferred their car to be black. Then a popular car reviewing publication began dolling out bad reviews if the car was in any color but black given the new research. Furthermore, they began advocating for the entire industry to only release cars in black, because a majority of people prefer it. Companies could also release cars in other slightly different shades of black, but ultimately, a variation of black is the only real option for a positive review. Given all of the bad reviews, car companies actually started doing it, given it was hurting their bottom line.

Now a high-end boutique car company comes along and releases a car in purple. Its niche audience really loves the new color, even though it isn't verified by any research. Regardless, the car publication gives it a bad review, predictably, but says if you spray paint it black, then it gets a full recommendation! They also go onto say, if you follow our testing methods, you will find that your preference actually IS black, you just aren't aware of it.

Do you see why that might be irritating to the niche audience of the purple car company?

Continuing on with my car analogy (I'm not even into cars, but a lot of people seem to be), using tubes in your audio gear is like driving a classic sports car. It isn't the best performing, it doesn't even have power steering, it gets horrible MPGs and overall is terrible inefficient. But it looks cool as hell and is still extremely fun to drive, compared to a sports car with contemporary engineering and technology.

While I agree with most of what you've said in this thread I do think the car analogy is very disanalogous given the preference is entirely based on sight while one of the issues in audio is that sight actually biases our evaluation of sound in a way that doesn't happen when we're blind. A better example would've been with something like wine, which we know is subject to similar sighted biases as audio.

 

[L0rdGwyn]

While I agree with most of what you've said in this thread I do think the car analogy is very disanalogous given the preference is entirely based on sight while one of the issues in audio is that sight actually biases our evaluation of sound in a way that doesn't happen when we're blind. A better example would've been with something like wine, which we know is subject to similar sighted biases as audio.

It's not perfect, but that's why it's an analogy. The sentiment is the same: people don't want to be prescribed preferences, even if it's based in market research.

 

[IAtaman]

Why would you be surprised if it's not audible? I mean, I posted another measurement in which the dip doesn't even exist and measures the same in that region as the Stealth. I think there are two issues in terms of if it's audible and then, if so, how much. I do think it's a mistake to assume it's as audible as the measurement suggests. As for "why not iron it out in the first place," that assumes that, one, it's easy to do or that, two, doing so won't introduce other (perhaps more audible) issues.

Are you arguing the dip does not exist, or the dip exists but is lower in level, or that a 6dB dip at 4KHz with Q=2 would be inaudible?

If you are arguing the dip does not exits because you shared a measurement that showed the same result as Stealth, the answer is simple : either your measurement is not correct or sth wonky is going on with Stealth. Amir's measurements, measurements Zach shared earlier and Techpowerup's measurements all show a dip. There are more measurements in squiq.link as well, and they are all showing the dip. The dip is there.

If you are arguing the dip is not as big: that is also unlikely because Amir's measurements, Zach's measurements and Techpowerup measurements all point to a 6dB dip which I estimate to be around Q=2 in size. Many others in squiq.link again show a simiar size dip - you can check yourself.

If you are arguing a 6dB Q=2 dip at 4KHz is not audible, you are effectively saying we can not detect a medium Q, 6dB difference in level at the frequency where our hearing is most sensitive. Q is too big for masking effects, 6dB is too large of a change and 4KHz is too sensitive - so no, that is very unlikely in my opinion. But don't take my word, put on your headphones, play a test track, dial in a -6dB Q=2 filter at 4KHz and see if you hear it or not. If you can not, go see a doctor.

 

[IAtaman]

I do plan on actually testing this myself the next time I use the Caldera. I'll be able to blindly switch a 4k EQ filter on and off. My prediction (which I'm open to being completely wrong about) is that if the 4k dip is audible it won't be nearly as severe as the measurement and overcorrecting for it will sound worse.

For the record, this is a new addition to your post which was not there when I was responding. I did not leave it out.

I agree, you will hear it and it will not be very severe. I did try it as well, it changes the tonality in a way it makes things a bit dull. It maybe affects the spatial qualities a bit? I am not sure about that one. And yes, it is not bad as such, but noticeable and certainly audible in AB test.

This is also a conjecture only but I can not imagine a large dip at 4K would be the natural way most people hear things normally. It might be a preference though.

Maybe try a headphone that does not have this dip to test. If the dip is the result of a cancellation in Caldera, EQ might not be that effective.

Edit note : Noticed Maiky76's EQ settings has this filter:

Filter 5: ON PK Fc 3703.41 Hz Gain 8.67 dB Q 3.23

So looks like the dip is deeper but skinnier, at least as per Amir's data. Maybe try with this settings to see how it affects the sound.

 

[Mulder]

I don't disagree with you, and I do believe in the scientific method and verifiability, which is why I suggested the Harman research should be verified before being put into practice or measured against as an industry standard.

Here's an analogy that maybe illustrates better my feeling - say Ford over the past five years did a significant amount of research on buyer preference of car color. What they found was 64% of car buyers preferred their car to be black. Then a popular car reviewing publication began dolling out bad reviews if the car was in any color but black given the new research. Furthermore, they began advocating for the entire industry to only release cars in black, because a majority of people prefer it. Companies could also release cars in other slightly different shades of black, but ultimately, a variation of black is the only real option for a positive review. Given all of the bad reviews, car companies actually started doing it, given it was hurting their bottom line.

Now a high-end boutique car company comes along and releases a car in purple. Its niche audience really loves the new color, even though it isn't verified by any research. Regardless, the car publication gives it a bad review, predictably, but says if you spray paint it black, then it gets a full recommendation! They also go onto say, if you follow our testing methods, you will find that your preference actually IS black, you just aren't aware of it.

Do you see why that might be irritating to the niche audience of the purple car company?

Continuing on with my car analogy (I'm not even into cars, but a lot of people seem to be), using tubes in your audio gear is like driving a classic sports car. It isn't the best performing, it doesn't even have power steering, it gets horrible MPGs and overall is terrible inefficient. But it looks cool as hell and is still extremely fun to drive, compared to a sports car with contemporary engineering and technology.

I agree with @solrage that your car analogy is a bit lame, but I think I see what you're trying to say.

I think you are mixing up different things. No one is saying you or anyone else can't like tube amps (or a purple car when following your analogy) If you're happy with how your tube amp sounds, no one can say you're wrong. The problem arises when you want to claim that your subjective tube-experience trumps measurement data. @amirm recommendations are based almost entirely on measurement data, which is the only objective way to assess the characteristics of a HiFi device. Everything else becomes subjective and individual. It does not mean that it is wrong for that individual, but it is not possible to generalize based on the individual subjective experience.

Then exactly how this data should be translated to the individual level is another question. For example, what is a reasonable level in terms of distortion? When do we start hearing it and what does it mean for our experience of the recordings we listen to? See for example the test of the 3e Audio 480-1-29A, TPA3255 class D amplifier module the other day, where exactly that question appeared in the discussion. Does the module cause audible IMD?

The Harman curve, which this entire discussion in the end is about, has a strength in that it is based on aggregated subjective preferences. It may not be the last word regarding standards, it may be adjusted in the future if more research is presented, but so it is with standards - they are not always set in stone. But most importantly - it's currently the only thing out there that resembles a standard. This does not mean that everyone will prefer headphones that follow the Harman, as the researchers behind the curve themselves point out, but it is the sound profile that is most likely to be liked by the most people. So, yes, it is reasonable that amirm uses the Harman curve as a reference, and no it is not wrong to prefer a headphone that deviates from the Harman. Incidentally, I myself daily use a headphone that deviates from Harman - OLLO S5X. A headphone that amirm also felt was in need of EQ to sound good.

 

[solrage]

Are you arguing the dip does not exist, or the dip exists but is lower in level, or that a 6dB dip at 4KHz with Q=2 would be inaudible?

If you are arguing the dip does not exits because you shared a measurement that showed the same result as Stealth, the answer is simple : either your measurement is not correct or sth wonky is going on with Stealth. Amir's measurements, measurements Zach shared earlier and Techpowerup's measurements all show a dip. There are more measurements in squiq.link as well, and they are all showing the dip. The dip is there.

If you are arguing the dip is not as big: that is also unlikely because Amir's measurements, Zach's measurements and Techpowerup measurements all point to a 6dB dip which I estimate to be around Q=2 in size. Many others in squiq.link again show a simiar size dip - you can check yourself.

If you are arguing a 6dB Q=2 dip at 4KHz is not audible, you are effectively saying we can not detect a medium Q, 6dB difference in level at the frequency where our hearing is most sensitive. Q is too big for masking effects, 6dB is too large of a change and 4KHz is too sensitive - so no, that is very unlikely in my opinion. But don't take my word, put on your headphones, play a test track, dial in a -6dB Q=2 filter at 4KHz and see if you hear it or not. If you can not, go see a doctor.

Not to get all Jordan Peterson but define "exist." The dip clearly exists in Amir's (and ZMF's) measurement, but not in the measurement I posted earlier in the thread. It's easy to simply write off the other measurement as "not correct," but that's just an assumption. Dips caused by driver/cavity modes (which Amir suggested earlier might be the cause) may not be audible OR nowhere near as audible as measurements suggest; we know that from psychoacoustics about room modes at higher frequencies. In general, I'm agnostic on the issue of whether the dip exists and how audible it is. If forced to wager I'd wager it exists, but isn't as audible as measurements suggest it should be.

For the record, this is a new addition to your post which was not there when I was responding. I did not leave it out.

I agree, you will hear it and it will not be very severe. I did try it as well, it changes the tonality in a way it makes things a bit dull. It maybe affects the spatial qualities a bit? I am not sure about that one. And yes, it is not bad as such, but noticeable and certainly audible in AB test.

This is also a conjecture only but I can not imagine a large dip at 4K would be the natural way most people hear things normally. It might be a preference though.

Maybe try a headphone that does not have this dip to test. If the dip is the result of a cancellation in Caldera, EQ might not be that effective.

Edit note : Noticed Maiky76's EQ settings has this filter:


So looks like the dip is deeper but skinnier, at least as per Amir's data. Maybe try with this settings to see how it affects the sound.

Yes, I did edit my post.

Curious, when you say you "tried it as well," what are you referring to? I could be wrongly assuming but I don't think you have the Caldera, so by "trying it" it seems you're referring to adding a similar dip in your own headphone? I'm certainly not claiming that doing such a thing would be inaudible. What I'm saying is that if the driver itself is (fairly) linear in that region, but driver/cavity modes are causing measurements to see a dip, that it may be inaudible or less audible do to how brains filter such modes at higher frequencies. We know this happens with room modes for loudspeakers.

 

[MayaTlab]

They tested for this:

"Four different DT770 units were measured, each repositioned three times. No outliers (e.g., leaking positions) were detected in these measurements."

They didn’t test for what I talked about. What they did in the paragraph you quoted is to look for outliers when measuring and then averaging the replicator headphones on their 5128. The response of the replicator headphones on the fixture, unlike the other headphones, isn’t shown in the article.

They did not control for inter-individual variation, which we know is likely to be an issue for the DT 770 Pro : https://www.rtings.com/headphones/1-5/graph/7913/consistency-l/beyerdynamic-dt-770-pro/440

And they did not test whether or not the transfer function between the ear simulator and the average in situ response for the test subjects is similar to the sort of headphones Harman used in their studies, purposefully chosen for being less sensitive to coupling issues (ex HD800). We don’t know exactly how the DT 770 Pro behaved in that regard, but Harman themselves, among others, found out that most passive closed back don’t transfer between ear simulators and the average of a cohort of listeners similarly to headphones with low sensitivity to coupling issues, and don’t “not do” so consistently.

In other words we have a less clear idea of what the tests subjects actually experienced and compared than if they had used something like the HD800.

There are good reasons Harman chose their replicator headphones specifically and did not pick any pair of headphones at random, and it’s all explained in the articles.

Have you read the paper or repeating other people's arguments?

As the French expression says, “c’est l’hôpital qui se fout de la charité”.

 

[fredristair]

I think if you are a speaker person you will impose your prejudices on headphone tuning. I'm a headphone person and I'm quite fine with a rolled off subbass (dynamics) and a flat subbass (planars). I listen to Susvara mostly or HD650, so something close to the Harmon is great, but I would never EQ the subbass and if that's what I was looking for I would listen to my speakers. Takes away from the clarity that great headphones excell at. If you put me in a room with some knobs to tweak the sound I would tune it to be something like a Susvara or HD650. EQ makes all my headphones sound worse and certainly I have the means to EQ them and have done so a great deal. I think the tuning on Caldera looks interesting and novel to me. I would expect it to sound great with rock music.

 

[L0rdGwyn]

I think you are mixing up different things. No one is saying you or anyone else can't like tube amps. If you're happy with how your tube amp sounds, no one can say you're wrong. The problem arises when you want to claim that your subjective tube-experience trumps measurement data. @amirm recommendations are based almost entirely on measurement data, which is the only objective way to assess the characteristics of a HiFi device. Everything else becomes subjective and individual. It does not mean that it is wrong for that individual, but it is not possible to generalize based on the individual subjective experience.

Yes, I know. As I've said, I've been on this forum longer than any of you, I know the drill, I've participated in these discussions multiple times. I never claimed tubes were objectively better, what I said was:

I think even the most ardent objectivists would be shocked at how good something like this can sound.

That does not mean it is objectively better, only that I think it sounds good. But even saying something uncontroversial like that here is a trigger, and people jumped all over it. Likewise, I said people are happy with the DIY amplifier design I released, not that it is objectively better. In fact, there are basic measurements in the design document, which of course show that is does not measure as well as a solid-state headphone amplifier from the likes of Topping, SMSL, etc. That is the nature of single-ended tube circuits that don't use high degrees of negative feedback.

Now I did say anecdotally that I have made designs that I felt were subjectively worse than equivalent circuits that measured worse. And I've offered to do a blind test to confirm or deny that.

The Harman curve, which this entire discussion in the end is about, has a strength in that it is based on aggregated subjective preferences. It may not be the last word regarding standards, it may be adjusted in the future if more research is presented, but so it is with standards - they are not always set in stone. But most importantly - it's currently the only thing out there that resembles a standard. This does not mean that everyone will prefer headphones that follow the Harman, as the researchers behind the curve themselves point out, but it is the sound profile that is most likely to be liked by the most people. So, yes, it is reasonable that amirm uses the Harman curve as a reference, and no it is not wrong to prefer a headphone that deviates from the Harman. Incidentally, I myself daily use a headphone that deviates from Harman - OLLO S5X. A headphone that amirm also felt was in need of EQ to sound good.

Those who disagree with that standard or who do not feel it is representative of their preferences are not going to buy into the idea of making it the defacto frequency response target of the industry, which is what is being pushed here. Even if the target is backed by sound research, people do not like being told what they should or should not like. The power of choice is highly valued. That is the essence of my very straightforward and easy to understand analogy.

I don't want to spend another day arguing on ASR, I've said all I've wanted to say and I'm sure most of you want to see me gone anyway so this will likely be my last treatise. I may respond to small posts, but I can't be writing essays all day again!

 

[IAtaman]

Curious, when you say you "tried it as well," what are you referring to? I could be wrongly assuming but I don't think you have the Caldera, so by "trying it" it seems you're referring to adding a similar dip in your own headphone? I'm certainly not claiming that doing such a thing would be inaudible. What I'm saying is that if the driver itself is (fairly) linear in that region, but driver/cavity modes are causing measurements to see a dip, that it may be inaudible or less audible do to how brains filter such modes at higher frequencies. We know this happens with room modes for loudspeakers.

Yes, I do not have the Caldera; when I say I tried it, I mean that I dialed in a filter and did AB to see if I can notice the difference using the headphones I have on my head at the moment.

I am not an expert on room modes, so I might be wrong here. I am aware Dr Toole famously said a brain and two ears is not the same thing as a microphone and I think he was referring to the filtering that is happening in our brains when the indirect sound reaches our ears - meaning that if we have a narrow high frequency peak caused by a reflection in the measurements, our brains might be ignoring it. I don't think this is a similar case as the reflection times would be too small and Q would be too wide for them to be perceived as echos and filtered out, plus they are dips, not peaks.

 

[solrage]

Those who disagree with that standard or who do not feel it is representative of their preferences are not going to buy into the idea of making it the defacto frequency response target of the industry, which is what is being pushed here. Even if the target is backed by sound research, people do not like being told what they should or should not like. The power of choice is highly valued. That is the essence of my very straightforward and easy to understand analogy.

Personally, I'm fine with Harman being considered a "standard." What I'm less fine with is assuming that everything that deviates from that standard no matter how or by how much is bad or flawed. I've gotten both very comfortable using EQ so that I use it for every headphone regardless of how close they are to Harman merely because of the variability in how music is actually mixed. I think everyone serious about audio should use EQ for that reason, as Harman isn't a panacea for bad mixing. With that in mind, things like flat (measured) bass isn't even a flaw because of how easy it is to EQ to taste and mixing. This isn't to say we can't look at some wild deviations and say they almost certainly sound like trash and there's no excuse/reason for them, but there needs to be more nuance than the simplistic "any deviation is bad." That's when we go from science to dogma.

 

[solderdude]

We don’t know exactly how the DT 770 Pro behaved in that regard, but Harman themselves, among others, found out that most passive closed back don’t transfer between ear simulators and the average of a cohort of listeners similarly to headphones with low sensitivity to coupling issues, and don’t “not do” so consistently.

Seal for DT770(32) and HD800S

 

[solderdude]

For the record, this is a new addition to your post which was not there when I was responding. I did not leave it out.

I agree, you will hear it and it will not be very severe. I did try it as well, it changes the tonality in a way it makes things a bit dull. It maybe affects the spatial qualities a bit? I am not sure about that one. And yes, it is not bad as such, but noticeable and certainly audible in AB test.

This is also a conjecture only but I can not imagine a large dip at 4K would be the natural way most people hear things normally. It might be a preference though.

Maybe try a headphone that does not have this dip to test. If the dip is the result of a cancellation in Caldera, EQ might not be that effective.

Edit note : Noticed Maiky76's EQ settings has this filter:


So looks like the dip is deeper but skinnier, at least as per Amir's data. Maybe try with this settings to see how it affects the sound.

Take any song.
Use a sound editor program.
Apply an EQ similar as the response of the Caldera, or just 4kHz at the described Q and amplitude (ensure the gain did not change).
Save that file.

Play both files using ABX tester and there you can objectively determine how audible what aspect (or aspects) are.
Can stop the 'this or isn't audible' right there but yes... is a lot of hassle but you gain some personal knowledge of audibility levels and don't need to read papers on it.

 

[solrage]

Yes, I do not have the Caldera; when I say I tried it, I mean that I dialed in a filter and did AB to see if I can notice the difference using the headphones I have on my head at the moment.

I am not an expert on room modes, so I might be wrong here. I am aware Dr Toole famously said a brain and two ears is not the same thing as a microphone and I think he was referring to the filtering that is happening in our brains when the indirect sound reaches our ears - meaning that if we have a narrow high frequency peak caused by a reflection in the measurements, our brains might be ignoring it. I don't think this is a similar case as the reflection times would be too small and Q would be too wide for them to be perceived as echos and filtered out, plus they are dips, not peaks.

That's what I thought, and I didn't mean to imply such a filter wouldn't be inaudible.

I don't recall if Floyd said it's only peaks that are filtered for rather than dips, nor how wide/narrow it must be to be filtered for. I only started this tangent because Amir himself said he suspected those dips could be due to driver/cavity modes. I then merely mentioned that if they're modes it's possible our brain would also filter (at least somewhat) for them. I don't know that's the case, all I know is that when I've listened to the Caldera I never noticed a massive dip there, and our ears are pretty sensitive to that present region. A massive dip there should result in instrument/vocals sounding quite distant. A subtle dip there could add a bit of spaciousness at the expense of immediacy.

 

[solrage]

Take any song.
Use a sound editor program.
Apply an EQ similar as the response of the Caldera, or just 4kHz at the described Q and amplitude (ensure the gain did not change).
Save that file.

Play both files using ABX tester and there you can objectively determine how audible what aspect (or aspects) are.
Can stop the 'this or isn't audible' right there but yes... is a lot of hassle but you gain some personal knowledge of audibility levels and don't need to read papers on it.

All this would show was that EQing a song will be audible, which nobody is disputing. The specific issue is if the higher frequency dips in the Caldera are caused by driver/cavity modes (Amir suggested this), how much of that is audible due to psychoacoustic filtering, which we know happens for room modes at higher frequencies. The only way to test that would be with an EQ on the Caldera itself.

 

[Mulder]

I never claimed tubes were objectively better, what I said was

I wasn't referring to you specifically. I took tube amplifiers only as an example to illustrate a principle, or rather a principled conflict between different approaches, i.e. between subjective and objective. I thought that was obvious.

Those who disagree with that standard or who do not feel it is representative of their preferences are not going to buy into the idea of making it the defacto frequency response target of the industry, which is what is being pushed here. Even if the target is backed by sound research, people do not like being told what they should or should not like. The power of choice is highly valued. That is the essence of my very straightforward and easy to understand analogy.

Your argument seems to shift. On the one hand, you claim that you do not attach much importance to the reliability of the Harman curve, or that it is not verified. In other words, the research is not solid enough to be regarded as reliable. Then you claim that even though it's solid, some people don't like the curve because...well, because they don't like it.

You object to the Harman curve being held up as a standard. But it is only one of several criteria. there are other criteria such as distortion, comfort and fit etc.. It seems that you worry that Harman will lead to a situation where freedom of choice disappears. There is little likelihood of that, I would think. Worse is if you prefer a standardless state, because there is no alternative standard. The non-standard morass is the normality in HiFi.
I also like freedom of choice, but I want to be able to make fairly rational choices.

In addition. The likelihood that further research would dramatically lead to the abandonment of the Harman curve is not very likely. Then it is more likely that it will be about minor adjustments. The Harman curve may not be verified by a third party, but Harman's researchers did not sit around and roll the dice. Regardless of whether one's own preferences match Harman's, it is a starting point that can be used for one's own evaluations.