Meze Empyrean 2 - Official Thread

This is the place to discuss all things to do with the Meze Empyrean 2.

Here’s the review:

Here are the measurements:

Note - I will be adding GRAS measurements once we have the DFHRTF for the KB5000 (we have the data, the calculation for it is still being worked on though).

B&K 5128 - Raw

B&K 5128 - Calibrated

Somewhat V-shaped tonality, kind of like an HD 800 S with added bass warmth. Overall its actually quite good with the hybrid pads, my only real complaint is that there’s a noticeable 6khz peak, and the upper bass tends to be a bit on the thick side of things.

Suede pads (B&K 5128 - calibrated)

With the suede pads, the bass lift is gone and its a brighter presentation in general. So for folks who want a somewhat leaner tuning, this is the way to go, but the suede pads may also be easier to EQ with just shelf filters since it also eliminates any muddiness in the upper bass from the hybrid pads.

Air gap (B&K 5128 - Calibrated)

Harmonic Distortion (90dB)

Harmonic Distortion (110dB)

Harmonic distortion is very good, showing only 2nd and 3rd order harmonic distortion at 110dB, and still below .2% for most of it. This is one where you will go deaf before you hear any harmonic distortion, and you can have all kinds of fun with bass boosts should you desire it.

EQ Profile:

Based on Hybrid pads

Make sure you adjust pre-gain to -6dB for this to avoid clipping! Or reduce digital volume.

Measurements after EQ:

Based on hybrid pads


The Empyrean 2 is a massive improvement over the previous models, but its also far more ‘normal’ for the tuning. It’s still V-shaped with extra bass and extra treble, but it’s less of a headphone that does things to your music compared to the original - its a lot more normal in a good way, depending on your preferences for the previous ones.

Subjective qualities are also good, its more about image separation and spaciousness than dynamics and contrast. Its still a step behind the ultra high end but I find it to be good enough, and in general the extra ear gain gives it much more clarity compared to the originals.

Overall, the Empyrean 2 is far more competitive as a product than the previous ones, and Meze should be commended for making these improvements!


Very cool - glad to see Meze sonically improved (arguably) the best built and most comfortable headphone around.

Regarding the measurements, and I’m sure I missed this explanation somewhere so I apologize, the DFHRTF line does not have a 10db or 12dB downward slope applied, like what you did with the preference target, correct? It’s just stock DFHRTF?

The calibration used for the compensated graph is just the flat DFHRTF of the measurement rig, with no tilt applied. The tilt is shown as the preference window (and you can see the 10dB slope there as well). The reason for this is to show the objective ear transfer function, which is equivalent to like… how something performs based on the anatomy of a human ear, and then also show the preference window for how most people like their headphones to sound.

1 Like

Graphs are too busy and confusing. Your Arya Organic graphs were nicer.

BTW this kinda looks like Hifiman tuning with a lot of bass + low mids if i’m seeing this right.

The new measurement paradigm will take time for people to learn, but the people need to learn! Preference boundaries are here to stay.

Although maybe folks prefer a light mode view?

And yes, there’s definitely a similarity there with the Arya and some of the other egg-shaped HiFiMAN headphones too!

Is this just a reference point moving forward or industry wide? Most reviewers going to this?

I say we standardize on a reference catalog of knob photos:

Learn it!


I had one of those. It was fun! Tho I don’t EQ, just find a headphone with a chain I enjoy.

1 Like

The preference bounds should be adopted by everyone - at least those who want to show a more complete picture of how a headphone performs. That doesn’t mean previous graphs are “wrong” though of course, it’s just that they provide a more limited viewpoint without the well-established preference groups from the research being reflected.

We’re happy to help folks adopt this, and some have already started doing so.

With regards to using the rig-specific HRTF calibration relative to tilt, this is becoming more widespread since it’s the only way to compare datasets across systems - Crin is already doing this for example with his 4620 data.

It may seem a bit complicated but once you start using it, it gets easier haha.


Ok. So will people like Jude use the new one as well?

These just seemed easier to read but maybe it’s just me (not that it’s white).

That’s entirely up to him, but keep in mind he doesn’t use targets - at least last I checked. I’m happy to chat with him about it and I think he’d agree that showing preference boundaries is better than the singular Harman target, but whether he adopts this approach or not I’m not sure.

With regards to the graphs you’ve linked, yeah I totally agree there’s more visual clarity there, and it may even seem easier to read but that’s also in part because that’s what folks are used to. With any big changes, it’s going to take time for people to get used to it.

If you haven’t seen it yet, I recommend watching the video I did about it, explaining the rationale:


Ok, sounds good. Appreciate it and not trying to irritate you, just curious. I like the old version and guess I’ll just adapt. Not new for me haha!


I’ll still post raw measurements, now that we have the preference boundaries for them built into the system - so that will hopefully help people who are used to the old method.


I really like this video!


Yes that would be correct. Dark mode for reading. Light mode for information.


Turns out the people prefer the dark side:


We’re all just living in darkness.


Can you please recommend a simpler 5 peak +2 shelf band EQ profile for RME ADI-2 that fixes the biggest deviations in the frequency range?

I don’t use the RME ADI-2, but you may want to use the suede pads and add a shelf filter to boost 100hz and below. For treble, downshelf above 4khz. You could also try just downshelfing 5-6khz too.

The linked video here explains things perfectly- now it all makes sense. I thought I had already watched it, but it must have slipped through the cracks.

I really like this new method of displaying measurements - keep up the good work gents!

1 Like

Pref boundaries on the raw measurement data is nice to see alongside the new chart style to help the transition. Just need to see our favorite headphones charted so that we can see a relative comparoson to the gear we know & love.

1 Like