ZMF Bokeh Headphone Review

[solderdude]

The Caldera reviewed here earlier was US $3.500. This is $1.100. That is a rather huge price difference. Maybe the Caldera was perhaps was a bit more polished, but still - why this big difference in price?

Own development custom made planar driver (development time also adds) v.s. sourced and much cheaper dynamic driver I guess ?

 

[Mulder]

Own development custom made planar driver (development time also adds) v.s. sourced and much cheaper dynamic driver I guess ?

If so, it is kind of ironic, since this "less" developed headphone seems to "measure" better. At least from an Harman perspective. Still, 1.100 is a lot for a headphone, and despite itś price it needs EQ to meet the Harman curve.

 

[taotone]

Nice review and fabulous headphones.

For those who want nothing more or less than a Harman-tuned response in an ugly plastic shell then of course there are better options.

There is space in the world for hand-made luxury products that deviate from a Harman tuning on a standard rig, especially considering HRTF, and individual preferences.

ZMF is very transparent and happy to co-operate so that you can get a good idea of what you are getting before you buy. Surely that is all we want? I hope to hear these at London Canjam in July and if they sound nice will likely purchase a set.

 

[solderdude]

If so, it is kind of ironic, since this "less" developed headphone seems to "measure" better. At least from an Harman perspective. Still, 1.100 is a lot for a headphone, and despite itś price it needs EQ to meet the Harman curve.

Only 1 or 2 OE headphones (both DCA) really meet the Harman curve on a specific fixture though so basically all headphones would need an EQ to meet that.

But yes... ZMF headphones ain't the cheapest ones around. They have a certain fanbase though. I don't own any though....
Compliance to a certain target in certain conditions on a specific fixture is not a guarantee one will like it though. There just is a high probability that most people will prefer the tonal balance aspect.

 

[AnyDave]

If so, it is kind of ironic, since this "less" developed headphone seems to "measure" better. At least from an Harman perspective. Still, 1.100 is a lot for a headphone, and despite itś price it needs EQ to meet the Harman curve.

For right or wrong, the Harman curve isn't our universal target curve. At the end of the day, it's a preference curve and we find that our customers do not have a unified preference. So we offer different models with different frequency responses. If possible, it's always better for users to listen first to see what works for them and what doesn't. It's extremely easy to listen with your eyes, but that's not always the most effective way of determining what you'll actually like. And that's not a dig at the Harman curve. It is a great tool in that it's finally created a reference curve that is understood well enough that it creates a useful starting point. I don't think anybody has accomplished that before.

Back when I used to work at Etymotic, we had an accuracy curve that was derived from the actual response of the human ear biology. Internally, we called it the KBN curve, as the study was done by Mead Killion (founder of Etymotic), Ellior Berger & Richard Nuss (Both of E-A-R- and later 3M). I'm having a hard time finding it online, but there is a whitepaper about it titled: "Diffuse field response of the ear." To a certain extent, I still consider it the corrent reference response as it is an objective target that denotes fidelity to the source. We always sort of took the tack that accuracy is the preferred starting point. It's generally easier to EQ out of accuracy than back into accuracy, particularly back then when most manufacturers didn't publish frequency response curves and most reviewers didn't have the means to properly test and publish them. On a personal level, I enjoyed a flat and accurate bass response, but also liked having variants with more low end for when I'm in the mood. I tended to favor low corner frequency bass boosts, which aligns reasonably well with Harman. In fact, I had personal prototypes for years before I was allowed to release them as it was not within the ethos of Etymotic.

But the ultimate point I'm working towards is that the world didn't universally prefer the accurate Etymotic response and it was never adopted as a universal target, regardless of it's fidelity. Even if you ignore the deep insertion fit of Etymotics, which clearly isn't for everybody, the overall response didn't suit everybody's needs. Often folks wanted more low bass, but sometimes folks wanted less midrange, or more treble, or frankly, for the primary resonance to be in a slightly different spot as the 2.7-2.8kHz average is just that, an average. I view Harman in a similar light. It's a subjective preference target curve that is incredibly useful as it will work for a lot of people, but it's not universal.

This is an extremely long-winded way of saying "different strokes for different folks" also applies to audio. And that's the approach we have at ZMF. For the most part, it's served us well as there's value in having a variety of options.

 

[SeasButter]

For such a close-back headphone, the measurement is fine.

 

[Curupira]

This is what I was referring to when I meant it was a bit "ironic" that also Bokeh fails when it comes to the adaption to the Harman curve - and just as Caldera need EQ to fix what Amir in the earlier test called errors.

Aside from the null resonance at 2.5khz, the deviations can be considered "taste". I don't work at ZMF but what I could gather from their fanbase is that they are not the most enthusiastic EQ users, and a lot of them own more than one ZMF headphone that is vastly different from the other in FR.

What I take from this, and I could be wrong, is that the company caters to collectors that dislike the "blank canvas" approach we value here at ASR on regards to Harman pinna compliance and low distortion on bass; rather, they prefer to own multiple instances of different headphones that play better on different genres (add this to the fact that a lot of "high end" enthusiasts are DAC and tube amp collectors and you get my meaning) or, in the case of owning a single HP, simply don't want a neutral headphone: each rendition of color in tuning can be seen as a different interpretation of music reproduction by the headphone designer, as if the transducer were a instrument itself rather than a medium.

You call can BS on all of that, call it counterproductive and for the most part I'd agree with you, but then again, we're talking about a boutique company that makes unique wooden headphones way past most countries minimal wage, so the audience is not the most pragmatic.

 

[Robbo99999]

Yep, one should not (and can not) EQ sharp nulls.
When sharp nulls are accompanied with an overall drop over a larger frequency band in that range then that wide-band drop should be addressed.

A sharp null on a sweep may not actually be a deep null (but often is a resonance) and may just be a little 'late'.
Below a CSD of the Bokeh (could be different pads) made by Marv (SBAF)

Here one can see that the deep null at 2.5kHz is not as deep as it appears but just a about 1ms delayed and is a resonance.
This delay is also visible in the GD:

High Q peaks usually are (somewhat) detrimental to sound and could be EQ'ed.

Interesting, I don't fully understand so I'll summarise, so you think the 2.5kHz sharp dip in the frequency response is because there's a resonance happening there that is somehow making that part of the frequency response arrive late vs other frequencies so it's being missed out in the frequency response sweep because of the "windowing time" in the software? (And you're saying it might not be heard as a dip because it's just arriving late in that section and so may not be a dip there in reality when listened to?) And which you say is shown by the group delay pic, but then why isn't there also a sharp dip in the bass frequency response at 80Hz given that there's a massive sharp dip there in the group delay? And how does a resonance make it arrive late? Overall I'm a bit sceptical of the things being talked about in this paragraph.

Either way though that massively sharp & large EQ filter at 2500Hz he used would not be good practice, re the ringing I mentioned.

 

[solderdude]

At 80Hz there is pad bounce (the pads absorbing and delayed returning of those frequencies) which is common in most OE headphones and is a well understood phenomenon.
It also has nothing to do with the (sharp) resonance which most likely is driver/acoustic resistor related.
It isn't visible in the FR because it is relatively too narrow (just a few Hz) and the frequency is lower so a few ms is not as severe at that frequency as it is for higher frequencies (phase)

And yes, clearly the 2,5kHz dip is a resonance and thus only appears as a deep null. Have seen this in other headphones as well (albeit not at the exact same frequency of course)

 

[Robbo99999]

At 80Hz there is pad bounce (the pads absorbing and delayed returning of those frequencies) which is common in most OE headphones and is a well understood phenomenon.
It also has nothing to do with the (sharp) resonance which most likely is driver/acoustic resistor related.
It isn't visible in the FR because it is relatively too narrow (just a few Hz) and the frequency is lower so a few ms is not as severe at that frequency as it is for higher frequencies (phase)

And yes, clearly the 2,5kHz dip is a resonance and thus only appears as a deep null. Have seen this in other headphones as well (albeit not at the exact same frequency of course)

Hmm, I don't fully understand why the 80Hz wouldn't show as a dip in the frequency response, but I'll take your word for it. Right, and so you're confirming that you don't think the 2.5kHz dip is real and in reality when you listen to that headphone there wouldn't be a dip there - I suppose in that case Amir shouldn't have EQ'd up that dip, so you're somewhat at odds to Amir's sonic experience of that region. I don't think he should have EQ'd up that sharp dip either, but for different reasons than you - I just think that would create "ringing" due to the large dB boosting over that very sharp narrow region (like I've said before). Also, either way, I don't see that 2.5kHz "phenomenon" (I won't call it a dip) as a positive for the headphone.

 

[AnyDave]

Hmm, I don't fully understand why the 80Hz wouldn't show as a dip in the frequency response, but I'll take your word for it. Right, and so you're confirming that you don't think the 2.5kHz dip is real and in reality when you listen to that headphone there wouldn't be a dip there - I suppose in that case Amir shouldn't have EQ'd up that dip, so you're somewhat at odds to Amir's sonic experience of that region. I don't think he should have EQ'd up that sharp dip either, but for different reasons than you - I just think that would create "ringing" due to the large dB boosting over that very sharp narrow region (like I've said before). Also, either way, I don't see that 2.5kHz "phenomenon" (I won't call it a dip) as a positive for the headphone.

Well, I measure a dip there as well. The question is really the audibility of relatively narrow-q dips vs wideband dips. There's nothing wrong with EQ'ing the dip if you prefer, but I think it'd be worth A-B'ing them to see if you can hear the difference. Now, a narrow-q peak is often a different story and more easily detectable. And, of course, wideband dips and peaks are often something to pay atention to.

There's a good paper by Bucklein titled: "The Audibility of Frequency Response Irregularities" that is worth checking out. If anybody has an AES membership, you can get it here:

E-library page - AES

AES E-Library ← Back to search

www.aes.org

An wanted dip isn't a positive. The question is really how much of a negative it is. I can't answer that for anybody, but again, there is value to listening pre and post EQ to see if the EQ improves it, is neutral, or makes it worse.

 

[solderdude]

The 2.5kHz dip is real, just not as deep as the measurement suggests.
The bigger elephant in the room are the general lower Q dips below and above 3kHz but even that some people would not mind but if one wants to EQ to Harman addressing those would be much more fruitful then the resonance (the null)

The sharp 2.5kHz peak of the filter also adds ringing (what you said) which may or may not make the already present ringing (it clearly is a resonance) worse or not.
Sharp filters are usually not a good idea anyway as any other copy might have one at a s lightly different frequency anyway.
For lowering peaks a sharp filter can be used but not for filling in nulls.

 

[Smaestro]

@Robbo99999 , it is very possible that Amir's EQ made it better for him despite there being no sharp dip.

The 'corrective EQ' that Amir put at 2.5 kHZ, for 8dB and with Q=9 is wider than the "dip" that he used it for. The "dip" has a much higher Q, at 2.3kHz and 2.7kHz it is flat again (Q=12?), whereas the EQ boost up to 2kHz and 3kHz.

Assuming there is no sharp dip at all (and group delay is below 1.4ms so that should be inaudible), this EQ brings that area more in line with Harman, a little above even.

It overshoots Harman by 1-2dB (hard to tell, based on the CSD) in the 2.5-3kHz area. That probably sounds reasonably good to him because in the bass he adds 3dB, an similar overshoot over the deviation from Harman of 1-2.5dB. So in result, bass and treble peaks are balanced, with a little bit more "V" in between.

 

[Robbo99999]

The 2.5kHz dip is real, just not as deep as the measurement suggests.
The bigger elephant in the room are the general lower Q dips below and above 3kHz but even that some people would not mind but if one wants to EQ to Harman addressing those would be much more fruitful then the resonance (the null)

The sharp 2.5kHz peak of the filter also adds ringing (what you said) which may or may not make the already present ringing (it clearly is a resonance) worse or not.
Sharp filters are usually not a good idea anyway as any other copy might have one at a s lightly different frequency anyway.
For lowering peaks a sharp filter can be used but not for filling in nulls.

Good, yes, we agree on the sharp 2.5kHz peak filter being being a bad idea.

@Robbo99999 , it is very possible that Amir's EQ made it better for him despite there being no sharp dip.

The 'corrective EQ' that Amir put at 2.5 kHZ, for 8dB and with Q=9 is wider than the "dip" that he used it for. The "dip" has a much higher Q, at 2.3kHz and 2.7kHz it is flat again (Q=12?), whereas the EQ boost up to 2kHz and 3kHz.

Assuming there is no sharp dip at all (and group delay is below 1.4ms so that should be inaudible), this EQ brings that area more in line with Harman, a little above even.

It overshoots Harman by 1-2dB (hard to tell, based on the CSD) in the 2.5-3kHz area. That probably sounds reasonably good to him because in the bass he adds 3dB, an similar overshoot over the deviation from Harman of 1-2.5dB. So in result, bass and treble peaks are balanced, with a little bit more "V" in between.

There's no getting around the fact that the 2.5kHz Peak Filter that Amir used is extremely sharp and extremely massive.

 

[SlothRock]

Love this review and that Zach and team stood behind their product and got this headphone over to Amir. I've had quite a few ZMF headphones and I've loved every one of them. While they don't comply with Harmon, it's obvious that is just a preference choice from Zach and squad. If he truly wanted, it shows he could easily produce a headphone that perfectly complies with Harmon but that's just not the ZMF sound signature, which is perfectly fine IMO. And in comparison to almost every other headphone maker, they keep pumping out high quality headphones with fantastic customer service to boot which should be praised in this hobby with some makers with a less-than-stellar record there.

 

[Smaestro]

@Robbo99999
I definitely agree.

 

[Jimster480]

Not sure I'd like how this one sounds, I'm not too much of a fan of dynamic drivers anymore...

 

[Mulder]

developed headphoneOnly 1 or 2 OE headphones (both DCA) really meet the Harman curve on a specific fixture though so basically all headphones would need an EQ to meet that.

Yes, that is true, and I myself also prefer a headphone whose frequency profile does not match Harman . So Yes, @AnyDave, I am perfectly fine with "different strokes for different folks" . My headphone of preference (OLLO S5X) didn't fare very well in Amir's review either. I can sometimes actually be a little disturbed by the fact that Amir in his tests tend to evaluate headphones that deviate from Harman as afflicted with "errors". Of course, it does not have to be a matter of "errors" for this particular reason.

However, my only point was, for those who really prefer the Harman profile - it is maybe a bit "ironic" that the less expensive and maybe "less developed" ZMF headphone offer a better coherence to Harman. Well, maybe that was a bad expression, since "ironic" can be perceived as very value-laden, but that was not really my intention. I didn't mean to express any opinion about the audio quality of either Caldera or Bokeh in any absolute terms.

My post was written with the earlier review on my mind. After the Caldera review followed a long thread (40 pages) where the relevance of the Harman curve was discussed. And as a result of that discussion - at least as I perceived it (I might be wrong) - ZMF would send another headphones to Amir that better corresponded to Harman, and they have done so. But still - it requires EQ to fully meet Harman, so I suppose for those who actually prefer headphones with the Harman profile, ZMF cannot offer a perfect headphone?

 

[solderdude]

Yep, since 'Harman tuning' is not Zach's target it is unlikely they will ever make one that 'complies' closely out of the box.
A good thing is that Zach offers some 'personal tuning options' such as inlays and pads but EQ/tone control would be more effective but is probably not an option for some of ZMF's owners.

ZMF do offer some 'well tuned' (but not Harman alike) models though.
After all, not every one belongs to 'the majority of people that prefer a certain tonal balance' so it is a good thing to cater for those too.

 

[TuteTibiImperes]

Aside from the null resonance at 2.5khz, the deviations can be considered "taste". I don't work at ZMF but what I could gather from their fanbase is that they are not the most enthusiastic EQ users, and a lot of them own more than one ZMF headphone that is vastly different from the other in FR.

What I take from this, and I could be wrong, is that the company caters to collectors that dislike the "blank canvas" approach we value here at ASR on regards to Harman pinna compliance and low distortion on bass; rather, they prefer to own multiple instances of different headphones that play better on different genres (add this to the fact that a lot of "high end" enthusiasts are DAC and tube amp collectors and you get my meaning) or, in the case of owning a single HP, simply don't want a neutral headphone: each rendition of color in tuning can be seen as a different interpretation of music reproduction by the headphone designer, as if the transducer were a instrument itself rather than a medium.

You call can BS on all of that, call it counterproductive and for the most part I'd agree with you, but then again, we're talking about a boutique company that makes unique wooden headphones way past most countries minimal wage, so the audience is not the most pragmatic.

For some of us there’s enjoyment in both.

I was fully onboard with getting the most transparent gear and EQing everything to Harman in the past, and that can absolutely sound excellent.

More recently I’ve embraced trying out a subjective experience where I’m not focused on transparent gear or worrying about EQ.

I have a subjective setup with a Schiit Yggdrasil OG and Folkvangr running very tubey NOS Western Electric tubes, and while I’m sure that setup measures very poorly it does sound amazing and very engaging with my ZMF Verite Closed and Auteur.

I also have an objective setup with an RME ADI-2 DAC and a Schiit Midgard, and using the Oratory1990 Harman EQ profiles with LCD-5 that’s a completely different and completely transparent and low-distortion experience that’s also very engaging.

I’ve been spending more time with my subjective setup lately because it’s newer to me and has a bit of that new toy appeal, but I also love the tinkering aspect of it.

Changing out the tubes in the Folkvangr can change the sound considerably (probably moreso than in most tube amps due to its design) and changing out the pads on the ZMF headphones can have a huge impact on the sound as well.

There’s something satisfying about finding that magical synergy of the right set of tubes and the right set of pads that just makes everything suddenly click and just sound transcendent.