I'm obviously a bit late to this party, having been the last stop on the US tour. (But thank you anyway @Andykong and @MezeTeam!) There have been plenty of unboxing videos, photos and impressions by now, so I'll just add some measurements, some brief thoughts and some thoughts on other reviewer's thoughts.

I had carefully avoided reading anything about this headphone until after I'd had a chance to hear it. I then looked at measurements, and finally went through internet reviews looking specifically for the 'cons'. So I'll go in that order...

I have to admit that until now I had not been a fan of Meze Audio's IEMs, but the Advar is a terrific headphone. My brief thoughts on it: beautifully small and ergonomic shells (whose build reminds me of the RHA CL2, except I massively prefer the tuning of the Advar). Fit and comfort are absolutely perfect for me. It has a good, punchy bass and an overall slightly warm tonality. But only slightly - its FR has a pretty balanced tilt with plenty of detail and treble energy (there may be some small risk of overcooking the highs on brighter-sounding recordings, but the right choice of eartip can fix this). There's some curious illusion of a wider sound stage that might be a result of those interesting horn-shaped vents. One thing to be aware of - because of those same horn vents, isolation isn't quite as good as that in other (vented) dynamic-driver IEMs.

Measurements stack up mostly as expected, including a flat impedance curve (@ ~32 Ohm) and exceptionally low levels of distortion, typical of good-quality dynamic drivers. The FR shows the well-extended treble and, perhaps, the slight possibility of 6 kHz sibilance with certain eartips. The bass sounds to me slightly more impactful than I might have anticipated from its graph (perhaps because of the comparatively recessed mids). Relative to Harman, the Advar has a bit less sub-bass, but fractionally more mid-bass. However, it doesn't have that off-sounding timbre that sometimes results from an overly-bloated mid-bass. The Advar has, to my ears, a really good low-end punch. Its impulse response is fast and clean with very little ringing, although the driver wiring was apparently reversed on both earbuds of this tour unit. I suspect this was just a minor manufacturing oversight. Meze probably only worry about matching polarity between L and R channels and aren't worried about absolute polarity, i.e., the potential for a 180 degree phase rotation. (I believe this effect would be utterly inaudible in all circumstances, but I'd be curious to know if anybody disagrees or has a counter-example?)

The following shows the Advar's FR using stock (Final Audio Type E) eartips and SpinFit Cp100 eartips. Measurements are from a GRAS RA0045 and the Harman target is shown here only as an anchor point (I don't claim that Harman should be your preferred FR or mine):


There are inexpensive dynamic driver IEMs (e.g. 7Hz Salnotes Zero and Moondrop Chu) that get closer to the Harman target than the Advar, but I find the Harman target to have a bit too much lower treble for extended listening. What's probably more relevant is how the Advar stacks up against its peers (more on this below):


Distortion is amongst the lowest levels we've ever measured in an IEM:

Its impedance curve is essentially ruler-flat, so there's no concern over device output impedance shifting the frequency response:


Impulse response is also excellent, but curiously inverted:


Lastly, I want to specifically address some of the 'cons' posted on various reviews throughout the interweb. (Lavish praise might have been OTT in some places too, but that's easier to forgive if people have limited experience with IEMs.) I love to see honest criticism of any product that genuinely deserves it, but I feel some of the critique here needs a bit of push-back:

1) 'The tuning won't work for everybody.'
This 'con' could be leveled against every headphone ever made. Even the Harman target only has a statistical preference rate of about 64%. The Advar is not really an outlier.
2) 'Its technical performance is sub-par.'
The suggestion here seems to be that the reviewer's brain can perfectly separate out the tonal aspects of a headphone and accurately rank all residual magnitude and phase errors. I know many reviewers repeat statements like this nowadays, but that doesn't mean it isn't utter bull$hit. There's simply no way to confirm whether claimed 'technical errors' aren't simply another manifestation of tuning (i.e. tonal differences) or even placebo. I suspect such sentiments originated from people trying to justify purchases of multi-thousand dollar IEMs that weren't tuned as well as cheaper IEMs like the Advar. (My very expensive IEM's tuning sounds off, but there must be a silver lining because the headphone manufacturer would never cynically over-charge for a badly-engineered product. What possible incentive would they have to do something like that?) It doesn't seem likely that any multi-BA IEM (with their typical phase and distortion errors) would have any kind of 'technical' superiority over the Advar. At 80 phon, harmonic distortion errors in the Advar are below the threshold of human audibility.
3) 'It doesn't come with a balanced cable.'
Actually, this particular tour unit did. But why do we have this continuing irrational obsession with balanced cables? Don't you have enough power from your single-ended output to permanently damage your hearing? Are you going to be connecting this to a PA system via a cable that stretches to the other side of a football field? Unless you answered yes to one of those two questions, there is absolutely zero benefit in using a balanced cable with this or any other IEM. Save your money for something that will actually make a difference in your life.
4) 'The shells are too small.'
I understand how large shells prevent you from getting a good, comfortable seal. But too small? In the worst case scenario of the limit as Δx->0, you're still left with an eartip? Which you're free to choose??
5) 'Short nozzles = shallow fit.'
The nozzles don't look all that long, but they angle in to my ear canals perfectly. For me the fit is definitely not what I'd call shallow. I guess this one might be fair warning for others though, because if you only manage a shallow fit, it's going to push those treble peaks to lower frequencies.
6) 'The imaging isn't "holographic" enough.'
That might be true, but this isn't a fault of the headphone. Accurate 3D spatial perception requires your brain to receive the amplitude and timing differences from your left and right ears, uniquely sculpted from your anatomy. For that you need to record on a dummy head that's a perfect anatomical replica of yourself or use an appropriate HRTF mapping. We can't expect recordings made on free-field mics for loudspeaker playback to give the exact HRTF response for your ears (including the necessary levels of crossfeed) when played back via the Advar. If that happened, there'd be something seriously wrong with your Advar. I understand the origin of these sorts of statements - the reviewer just got it from somebody else and passed it on. The same thing used to happen with diphtheria, and that was also a bad idea.
7) 'It sounds different with different eartips.'
So do all IEMs. Eartips gives you some extra tuning options. The statement is correct, but I would consider this a pro, not a con.

If there are tweaks I might make to the Advar's tuning, these would be relatively minor. Out of the box, without needing to do anything more than find an appropriate pair of eartips, they sound fantastic. To my taste, they're up there with the port-modded Beyerdynamic Xelento and Sennheiser IE600 as some of the best headphones you can buy at any price. I consider the Xelentos and IE600 to be the Advar's closest relatives, being similarly-tuned, small, single dynamic driver IEMs. (The AAW W900 and Vision Ears VE8 both have similar tunings to the Advar - at least up to 10 kHz where the VE8 then rolls off abruptly. But both of these are multi-driver IEMs with much larger shells that are always going to be potentially problematic with fit.)
The Xelento is a difficult recommendation because there are now a lot of fakes out there that look very authentic (but don't sound as good) and the Xelentos also require a port mod to sound their best, otherwise their mid-bass is a little bloated. This leaves the Sennheiser IE600 as the Advar's major competitor. The Advar has slightly poorer isolation compared to the Sennheiser IE600, but I find the Advar's sound signature a bit more enjoyable. Both are class-leading IEMs though, and if I were forced to pick between them it would honestly be like Sophie's choice.

Summary

I can highly and unequivocally recommend the Meze Audio Advar. It has a combination of small, ergonomic, premium-looking shells with a fun, euphonic tuning that is inexplicably and inexcusably rare. FWIW, I enjoyed the Advar enough to buy them.

Disclaimers & Background

I hereby acknowledge myself for having supplied me with a Tanchjim Darling that I can keep after this review I made a rather unusual decision to purchase this IEM blind, based on design alone. I was pretty sure this would fit my ears without any issues; the sound was a very pleasant and unexpected surprise.

Now to explain my rating... It's tough to find any perfect product, and I don't claim the Darling is perfect, but it's so close I'd be nit-picking not to round it up to 5.0.

Build and Fit

The Darling is a hybrid IEM with a single dynamic driver and two balanced-armature drivers. The BA drivers sit forward in the shell, firing directly down the nozzle, which likely contribute to its excellent treble extension. While dynamic driver (and dynamic-driver hybrid) designs usually rely on external vent ports for tuning, the Darling doesn't. The result is a fantastic (Etymotic-like) level of external noise isolation. Sound-stage width collapses slightly as a result, but it's hardly a fault of the headphone if users choose to play back music recorded for speakers via deeply-inserted IEMs.

Sound and Comparisons

The Darling is relatively easy to drive and fairly insensitive to source output impedance. It's still worth noting that increasing source output impedance does tend to increase those mid-range and treble peaks, so I'd recommend a quality source with as low of an output impedance as possible. (A Hugo 2 is awesome, but a Shanling M0 or FiiO M3 Pro also work great.) I also strongly recommend the use of foam eartips, as these also help tame the treble. I found the best-sounding eartips for these to be the small-size Comply Comfort 500 tips. The Darling has notches on its nozzles to allow shallower or deeper insertion. Unusually (for me) I found I preferred the shallower-insertion option here:


So how does the Darling compare against the very best headphones out there?

The KSE1500 is probably the most resolving IEM you can currently buy. Theoretically, that comes at the price of some shift in timbre, as it moves much of the mid-range energy into the ~10 kHz region. Its main issue is really just portability as it requires a dedicated electrostatic amp, and many folks in the KSE thread have been hunting for a KSE1500-equivalent that doesn't require such an amp.


My favorite eartips for the KSE1500 are the double-flange SpinFit Cp240 tips. They're not the world's most comfortable eartips, but they help to tame the treble. Even with these eartips, the KSE1500 still has a lot of energy in the ~10+ kHz region. With a low output-impedance source and foam eartips on the Tanchjim Darling (all bets are off if you skimp on either one of these), it's honestly difficult to say that you're missing anything with the Darling. In fact, the Darling is closer to the Harman target and so ought to (theoretically) have a wider appeal than the KSE1500.

The Shure SE846 is still a popular IEM, and for good reason. Despite its age, it's still one of the best-sounding IEMs you can buy. Comparisons below use the combination of white filters with Cp240 eartips:


The Darling drifts slightly over the Harman target around 3 kHz, while the SE846 is a little further under. The Darling has better treble extension, which has often been noted as a weak point of the SE846.

Another fantastic TOTL earphone is the Final Audio A8000. This is a single dynamic driver IEM that leans a little bit bright with silicone tips, but sounds very good with foam tips. The cables on the Darling and A8000 are rather similar - both shiny silver with a transparent 3.5 mm plug on the A8000 and transparent mmcx connectors on the Darling.


The two frequency responses are quite similar, with the major difference being a peak around 6 kHz in the A8000, and a dip at the same location in the Darling, leaving the Darling somewhat less prone to sibilance.

Summary

It's rare to find an IEM this good at any price. The Darling easily competes with, and in some ways exceeds, many of the current best IEMs. It should make us all seriously consider whether we really need those $3000 Beryllium-coated electrostatic, vibranium-encrusted e-tweeters.

More details and comparisons available here:

https://www.hypethesonics.com/tanchjim-darling-minihype/

What to do when stuck at home trying to avoid coronavirus? I know, I'll write another review...

I'll skip the photos and the unboxing video, because you folks already know what the A8000 looks like and I have zero photography/videography skills anyway. Let's get to the important stuff.

I should first point out that I can metaphorically rip any headphone to shreds given enough time and blank paper. The A8000, like every other headphone, has its faults. That being said, the A8000 doesn't have many faults, and the issues I have with it are relatively minor. Let's get those points out of the way first.

1) The A8000 is an undeniably bright-sounding IEM. Its characteristic brightness comes from a series of treble peaks corresponding to driver and ear canal resonances. Is it so bright that it's intolerable? I don't think so, but the answer to this is going to depend on the insertion depth you can achieve (more on this later), your overall sensitivity to treble, the frequency content of the recorded/mixed/mastered source material and (to a lesser extent, given the flat impedance curve of the A8000), the DAC and amp.

With the A8000, my preferences mainly shift with the genre and spectral content of the recording. Certain female vocals (and Jon Anderson) can sometimes sound a little peaky. "Open Your Eyes" - the quite-listenable title track from the otherwise awful 1997 Yes album of the same name - is too bright for my liking with the A8000. Whereas the entire Tears for Fears catalogue (especially that 96/24 Steven Wilson remastered Songs from the Big Chair) sounds amazing on these headphones. There's something about the A8000 that makes it quite a special and unique-sounding headphone.

Here's a comparison of the average FR (from L & R channels) measured with two different types of GRAS coupler (RA0401 measurements courtesy of @jude):


When I first measured the A8000 (in my review of the FiR Audio M5:
https://www.head-fi.org/showcase/fir-audio-m5.23861/review/23350/), I saw the primary ear canal resonance peaks at around 7.5 kHz and 11 kHz. But when I did a frequency sweep, I heard them higher than this - closer to 8.5 kHz and 11.5 kHz. Sure enough, with a slightly deeper insertion in the coupler, these newer measurements (above) closely match what I hear. Note that the standard RA0045 coupler matches what I hear - the RA0401 coupler is designed to massively damp the ear canal resonance - and it works as advertised. Yay(?) So, the RA0401 is a great device for designers wanting to specifically isolate/identify ear canal and driver resonances, but it's not very useful when you want to actually know how an IEM would sound in your ears. I don't trust the RA0401 to faithfully represent what I hear. IMHO, you shouldn't either. (Thankfully, it looks like headfi will be moving away from using the RA0401 for future measurements: https://www.head-fi.org/threads/brüel-kjærs-evolution-of-hearing-simulation-part-2-coming-soon.928514/post-15533359.)

The RA0045-measured spectrum shows that the A8000 still follows the Harman target closely in the low frequencies, but it doesn't hug that target quite as well in the treble and has a few small peaks that are audible with a frequency sweep. I had some communication with Final Audio about this and they told me these resonance peaks were an intentional design choice, because although they could remove them, they felt the headphone sounded boring without them. I believe that's absolutely true. Any downward adjustments to those treble peaks would need to be very subtle so as not to risk destroying the fairly addictive sound-signature that the current A8000 design has. This is definitely a case of being careful what you wish for.

2) As mentioned above, the A8000's shell has some razor-sharp edges - as in, sharp enough to use it as a cutting tool - this edge, and this corner, in particular:



Perhaps the sharp edges won't come into contact with your ears? They don't seem to bother me, but I read one comment from somebody on the forum who had cut their ear with the A8000. A slightly rounded/chamfered edge could still have looked nice and would perhaps have removed that potential problem?

Having got those two issues (slightly bright-leaning sound signature and sharp edges) out of the way, everything else I have to say about the A8000 is positive.

Channel Matching

Channel balance is good:


Fit

When I put the A8000 into my ears, the top part of the IEM body just naturally rotates out a bit under its own weight, leaving the face-plate at a bit of an angle (roughly 45-degrees to the vertical) and this naturally pushes the IEM farther into my ear canal, which pushes up the frequency of those resonance peaks. (In my experience, most IEMs sound better with deeper insertion for exactly that reason.) I'm not entirely sure how the A8000 is supposed to fit, but when I asked Final Audio, they just said "whatever works". Well, for me it works I would never have anticipated this, give the weird, non-ergonomic, boxy/sharp-edged-looking shape of these, but the A8000 is possibly the best, most comfortable-fitting IEM I've ever used. It's quick and easy to insert and it isn't ever fighting to get out of my ears. I can't tell you how much of a rare find that is for me

I should also mention something about isolation. The A8000 is vented, so it won't isolate as well as an Etymotic. However, surprisingly, I found its isolation to be actually quite good -FWIW, slightly better even than the Sony WF-1000XM3 earbuds with their active noise cancellation on.

Impulse Response

The A8000's impulse response is not quite as fast as that of the KSE1500 (but nothing else is in my experience). Its speed is more at the level of the dynamic driver in the Beyerdynamic Xelento, however, it has better damping, which I suspect has something to do with the Beryllium used in the A8000's driver:


Square wave response is reasonable. I've seen a lot worse:


Distortion

I'm not entirely convinced that harmonic distortion levels are all that important, but for those that care about this sort of thing, they're extremely low on the A8000:


The only other headphones I've measured with a distortion this low are the Beyerdynamic Xelentos, and honestly, it's pretty hard to pick a winner between them in that regard (see graph below - note this particular measurement was at 80 dB @ 500 Hz). I suspect the THD is so low in both these IEMs that it's hit the floor of my measuring equipment (likely limited by the THD from the coupler microphones):


There's another point about the THD worth mentioning. Despite the heavy ear-canal-resonance damping in the GRAS RA0401 coupler (which should produce artificially-low levels of THD), the THD percentage levels (normalized w.r.t. the fundamental) I measured at 90 dB on a GRAS RA0045 coupler were notably lower than those measured by @jude at the same SPLs:



The reason for this may be that Jude's distortion levels seem to have been taken directly from a sweep measurement. (Note the better agreement above with my standard sweep THD measurement. This was done at 94 dB @ 500 Hz, but it's close enough to show the effect. The stepped-sine measurement was 90 dB @ 500 Hz.) For harmonic distortion measurement it's more accurate to use a stepped-sine approach. This uses fewer steps (hence the reduced frequency resolution), but is able to more accurately pick out the harmonic components at each frequency bin. The settings used above are my standard approach for stepped sine: 131072 FFT length, rectangular window, 4 sliding averages with a 50% overlap and 6 points per octave. Anyway, the bottom line is, the A8000 has incredibly low distortion.

Impedance as a Function of Frequency

The A8000 has a beautifully-flat impedance curve:


What this means in practice is that the frequency-response curve shown earlier can be achieved with virtually any source, even your phone. And even with a HiBy R6

Comparisons

Here are some thoughts on how the A8000 stacks up against what I consider to be its most worthy competitors (my current favorite selection of IEMs):


Final Audio Thoughts

The bottom line is the A8000 is a very nice-sounding headphone. To respond to some negative comments I've seen elsewhere... If you're worried about the weird shape and whether they'd fit you or not, I think you're likely to be pleasantly surprised; at least, they fit me amazingly well. If you're worried about the solder being visible on the semi-transparent 3.5mm plug, don't be. It's just a design aesthetic (like being able to see the PCB on a Hugo 2) and actually shows a very neat soldering job. If you're worried about it lacking in bass - don't be; it isn't. The low frequencies follow the Harman target closer than any other headphone I've seen and it has a fantastic sub-bass without also dumping tons of energy in the mid-bass. The low-end of the A8000 is pretty much flawless to my ears. I think the reason that some have commented on a "lack of bass" is simply because it tends to lean a little bright with those treble peaks, and that shifts the overall emphasis away from the bass. That potential brightness is probably the only real caveat with the A8000, but your experience in this regard is going to depend on other factors such as source material and whether you can get a deep-enough insertion to push the ear-canal resonance peaks out of the regions that might bother you. For me, they sound bright, but rarely ever come across as too bright. They are certainly one of the best IEMs I've heard in recent years - easily up there in my top 5

Ouch, that's a lot of cons. Why would I give 4.5 stars to a headphone with so many negatives? Well, because the M5 can sound absolutely fantastic. I've tried one pair of M5s that were possibly the most enjoyable in-ear monitors I've ever heard. To my ears they sounded extremely natural, with good extension at both ends - not overpowering, but just enough of a lift at the extremes that they never sounded boring.

If you purchase the universal model in the US, you can also customise your own cable on the FiR Audio website, which is pretty cool. I'd recommend memory wire in the ear hooks because of possible fit issues (more on that later). The stock cables are soft, flexible, have very low microphonics and seem to be high quality.

Some more thoughts on those cons...

Price: If there's a company out there who feels their R&D talents justify a massive hourly rate, I guess they're free to set a commensurate price for their headphones and the market will decide. But the parts that go into these headphones wouldn't seem to justify a nearly $3000 sticker price. I guess we should just be patient though... in 10 years time when there are $50,000 IEMs on the market, $2800 will look like a ChiFi bargain.

Isolation: The M5 is vented, so it doesn't offer much passive noise isolation. If sound isolation is your highest priority, you'd still be better off with an Etymotic.

Channel Matching: It was reasonable - not 100% perfect on the first demo pair I first tried - but reasonable. Note, all M5 measurements here were made on a standard 711 coupler with Cp100 SpinFit eartips. These graphs are all uncompensated:

Despite the small channel imbalance, these still sounded wonderful. After I made the mistake of pointing out the channel imbalance on this particular demo pair, Bogdan (FiR Audio owner/CEO) re-tuned them. The result was then near-perfect channel matching, but a disastrously-different FR, with a much heavier mid-bass, and a dull, muddy and pretty lifeless sound. So I guess the moral of this story is be careful what you wish for? ¯\_(ツ)_/¯

High unit variance: I expect subjective impressions of the M5 will have big variations. By which I mean even more variation than is normal for headfi. This is because of a couple of possibly not-so-good points, namely, low/variable impedance and very noticeable variations in production units. As a result, subjective opinions might not be all that helpful here. Showing a frequency-response graph gets a bit tricky too because I've heard three models that all sounded quite different:

Are you ready to roll the dice?!

One positive point worth mentioning here though. In certain IEMs, foam tips can massively damp the treble. Because the M5's bores are fairly wide, I didn't notice that much difference between silicone and foam eartips:

The M5 comes with its own proprietary brand of foam eartip. It looks like Comply, but apparently it isn't.

For whatever reason, big-name-brand companies seem to do better with QC in terms of tight manufacturing tolerances. Unit variance is possibly the Achilles' heel of the M5. From model to model, its FR appears to be all over the map. @crinacle also has several measurements for the M5 on his website and none of them match either. (Note @crinacle also shows a sizable FR spread for the other FiR Audio IEM models.) Bogdan suggested couplers and eartips might be responsible for the variations in measurements, but I don't entirely agree. I have a fairly careful procedure and all my measurements were made with the same eartips, same insertion depth, same coupler, same mic, and are 100% repeatable. From what I've seen of @crinacle's work, his measurements also seem to be carefully done and are usually pretty consistent with mine. (My measurements were made on a GRAS RA0045/40AO coupler/mic, but I also have the same coupler that @crinacle has, and measurements from my various 711 couplers match very closely for FR.) These unit variations with the M5 are real and easily audible.

Fairly high levels of 3rd harmonic distortion: At 80 dB, it's ok. Typical of what I usually see in other balanced-armature IEMs:

At 94 dB, the third-order harmonic distortion got so high in the mid-range that my measuring software quit (by default, REW stops if distortion exceeds 1%):

Does this matter? Probably not. This isn't random noise or intermodulation distortion, but harmonic distortion - in other words, harmonics that could/would exist in nature. In theory, this could change the timbre of a sine wave, but in actual music the dominant factor is the underlying frequency response, given that the harmonics of the instrument(s) are already present in the recording and the perceived FR of the headphone is almost certainly going to be deviating from flat by way more than that THD percentage anyway. Consider the irony of chasing a 0.0001% THD and then plugging those same headphones into a tube amp

Impulse Response: The M5's impulse response is fairly typical of balanced-armature headphones - rise/fall time is not as fast as electrostats like the KSE1500, but settle time is faster. It's also faster than a dynamic driver like the Xelento, and much faster than the planar driver in the RHA CL2:




Does any of this matter? Possibly not. Although it's intrinsically interesting, I haven't seen much evidence to suggest that we have an urgent need to be able to resolve impulses <50 microseconds apart in real music.

Square-Wave Response:

Ok, this doesn't look much like a square wave, but again I'm not convinced this has much bearing on real music.

Very Low (and Variable) Impedance: The quoted specs say 6.8 Ohm, but this is the impedance I measured, which was pretty consistent on both channels of two different sets of M5s:

The impedance is 6.8 Ohm only once you get below 100 Hz. It's way lower than that at frequencies above this, and drops below 2 Ohm in the mid-range. I've never seen a headphone with an impedance this low. Together with its variation with frequency, this definitely could have an audible effect. How many people have source devices with an output impedance lower than 0.25 Ohm? My guess is not many. I've seen mention of LID (linear impedance device?) in some other M5 reviews, but can find no mention of this on FiR Audio's website. If the M5 does have some magical device to prevent shifts in FR with varying output impedance, it doesn't seem to be working very well. Here are measurements of the exact same headphone (without shifting anything in the coupler) being driven by an RME Babyface Pro FS and an A&K Sp2000 which have output impedances of 0.1 Ohm and 1.0 Ohm, respectively:

Both these output impedances are already pretty low, but there's still more than a 2 dB shift in the mid-range. It seems safe to expect the FR to shift noticeably when moving to/from devices with larger (>=1 Ohm) output impedances.

Sensitivity: I couldn't find any official specs listed for this, but I measured it as approximately 110 dB/mW @ 1 kHz. That should mean that the M5 is unlikely to exhibit any noticeable hiss. And even if it does, that wouldn't be the fault of the M5 and you really ought to get yourself an amp with a lower noise floor

Ergonomics: If you squeeze lots of drivers into one IEM shell, you're probably not going to get something as nice and tiny as a Xelento. But why not at least make them ear-shaped and angle the nozzle up into the ear canals properly?
The fit of the M5 is ok. It's not the horrible torture that is the Campfire Audio Andromeda, but it could have been so much better - at least for my ears. I need to rotate the M5 back a bit farther and angle them out in order to get a good seal, and as a result, I've had to use a couple of funky bends in the ear-hook memory wire in order to get the cable to stay behind my ears, as the natural tendency, given the fit of the IEM, is for the wire to end up hanging next to, but not behind, my ears. If any of that makes sense?

Proprietary Connectors: The M5 comes with a pretty decent stock cable which you probably won't want to change - which is just as well, because you won't easily be able to. The M5 uses FiR Audio's own proprietary RCX connectors. I understand not everybody likes MMCX connectors, but, on the other hand, a proprietary connector makes it difficult/impossible to ever switch out the stock cable. I like the M5's connectors, but I can't see these catching on as a new universal standard anytime soon. And in the meantime, none of my existing cables are interchangeable with those on the M5. If you're handy with a soldering iron, FiR Audio sell their RCX connectors for $30 a pair; if not, you're probably not going to be doing much cable rolling.

Comparisons:

Conclusions

This is the most difficult part of my review. Can I recommend the M5? I don't know. Most of the negative issues listed here should be relatively minor to most people. The main issues are: 1) whether you can get the M5 to seal and stay in your ears, and 2) if you're lucky with the tuning on the unit you receive.

Unit variance with the M5 can be huge, so even if you've heard a demo pair you like, you're still going to be taking a bit of a gamble on a purchase. Compound this with the uncertainties in tuning any CIEM, and I could never recommend a custom M5. (But then I don't recommend a custom anything.) I'd strongly recommend prospective universal M5 buyers to contact FiR Audio and ask to audition the exact pair they intend to buy. If that's not possible, I'm not sure I could recommend the M5. I've now heard three M5 models: one sounded phenomenal, one sounded good, and one sounded pretty muddy. That kind of loose manufacturing tolerance is not ideal at any price, but it's a poor show at $2800. In fairness to FiR Audio, unit variance of other headphones probably isn't well documented. It's common knowledge that reviewers tend to get A-stock, and actual buyers are often left with somewhat lesser-sounding production units - and in most cases, those buyers probably never ever find out. This might even be the root cause of many disagreements on headfi, where two people think they're discussing the same headphone If you can demo an M5 with the option to buy the exact unit that you heard and liked, then I would go for it

Update (Feb. 2021): Comparisons against the M2, M3 and M4 can be found here: https://www.hypethesonics.com/fir_audio_mini-hype/

Acknowledgements and Disclaimers

I'd like to sincerely thank Colum and the rest of the good people at RHA for kindly allowing me to be part of the North American CL2 tour. The CL2 used for this review was loaned to me and was passed on to the next reviewer at the end of my week-long review period.

Any impressions from the measurements in this post are inferred at the reader's own risk. No measurement will tell you exactly how you'll hear, as our hearing is unique to each of us. Measurements presented here are raw (uncompensated) and should only be used to make relative comparisons and not direct comparisons to other measurement rigs.

Initial Impressions

RHA are a relatively young company, but with some good pedigree behind them. You might know them from their very underrated T20(i) IEMs. At least, you should know them for that reason - the T20s are great headphones that sound way better than their price tag would suggest. So I was super-excited at the prospect of seeing and hearing their new CL2 - the world's first proper planar-magnetic in-ear monitors. (Ok, I guess, technically, Audeze got there first with their planar-magnetic TIE Fighter head-accessories, but Audeze's iSine/LCDi4 planars aren't isolating at all and they're huge and they look ridiculous. RHA have managed to engineer a small, isolating planar IEM that looks infinitely cooler than the iSines or the LCDi4.) Rather than just add more off-the-shelf balanced-armatures into yet another multi-BA IEM, RHA have really innovated something different and something special here - a single planar-magnetic driver in a small, ergonomic, traditional IEM shell:



Right out of the gate, things are looking good - the packaging and presentation look premium and they come with an impressive set of accessories:



Fit, seal, isolation, comfort, profile and aesthetics are all top-notch with the CL2. There's really nothing I can fault here. Kudos to RHA for understanding we don't all want to have something the size of golf balls sticking out the sides of our heads. Also, whether by luck or thoughtful design, the nozzles are precisely angled for my ear canals, so that the buds themselves sit perfectly in my concha bowls. That's a surprisingly rare find for me.

The accessories included are impressive and are actually useful. Included are a variety of silicone and (Comply) foam eartips (I had no issues achieving a good fit and seal with either type), a 3.5 mm single-ended and a 2.5 mm balanced cable and a Bluetooth neck-band. The cables appear to be of high quality and though I don't always enjoy memory wire, I had no particular issues with these. So far, so good.



Sound Impressions and Measurements

Now for the not-so-good news. In the first few seconds of listening to this headphone I knew something was off. It sounded like the all-too-common lower-treble peak, so I switched from silicone to foam tips and that helped a bit, but something still sounded off. So I then put the CL2 on my measurement rig. Yikes RHA - that's a very unconventional tuning! The following measurements fairly closely match measurements from RHA that I was able to find online. My measurements were made using a 711-compliant coupler using REW and are presented here raw, with no added compensation other than that for the mic and sound-card. It appears that RHA's measurements were also raw:



There are some discrepancies in the high and low frequency extremes, and I give a list of possible reasons for that, along with measurement procedures and lots more disclaimers in this thread: https://www.head-fi.org/threads/audio-measurements-on-a-headfi-budget.893084/. The discrepancies at the lower frequencies are most likely due to my coupler & mic. I suspect the differences around 9 kHz are most likely due to RHA's out-sourced measurements using a newer GRAS hi-res coupler, which imposes a strong (but not necessarily physical) damping at the half-wave resonant frequency of the ear canal. Listening with foam tips to a frequency sweep with a tone generator gave the impression of a more even response between 6-10 kHz, i.e., most likely some kind of average of the two measurements. YMMV in this regard, because of your own ear canal resonances and impedances. However, the most significant feature of the FR is in close agreement in both curves, and this is the large bump in energy between about 2-5 kHz, followed by a substantial drop in the treble. There's always the possibility of unit variation, but the channel matching between L and R buds was fair and both show the same effect:



I contacted RHA to ask about measurement tolerances and whether they felt the frequency-response in this region was indeed indicative of all CL2s, and not just this particular tour unit. This was their reply:

The graph we provided is from the golden sample, which is kept in our manufacturing centre for comparison versus units from the line. Our tolerances are to +/-1db; the above should be a fair representation of anything coming off the line (and they are tested to this standard).

Click to expand...

So I think this rules out any specific defect in my tour unit in that regard. The CL2 really is supposed to have that bump between 2-5 kHz. Hmmm... Now, eartips can make a difference. As shown below, foam tends to roll off the treble, which was why I switched to the foam tips - for me, the CL2 sounds better with foam tips (I suspect some of you may also find the ~9 kHz and ~13 kHz peaks with the silicone tips a little strident). For this reason, all other measurements here were made only with Comply foam tips. The following figure shows the differences with the two types of eartip:



As you can see, foam eartips have the most pronounced effect in the treble region. They do lower the mid-range bump too, but only by a pretty small margin. The reason for the tuning in the CL2 is very likely the same reason behind diffuse-field compensation (explained more at that http link above). Since birth, our brains have had to compensate for the fact that, because of our ear canal resonances, we receive a boost at our eardrums in the 3-4 kHz region. Ironically, we often get a dip at ~4 kHz when we start to lose our hearing because those cilia have taken a heavier beating during our lives - so older folk might actually prefer the CL2's tuning. But the theory for some level of mid-range boost applies to everybody: when you take away the ear canal resonance mechanism by shoving something deep into your canals, you want to account for that boost in the IEM tuning. Or do you...?

Our brains are amazing at adapting to skewed input data. I'm 99% convinced that 99% of burn-in that people hear is nothing more than our brains adapting to skewed input. (There's a classic example of this here: https://www.theguardian.com/education/2012/nov/12/improbable-research-seeing-upside-down.) I have measured differences before and after burn-in, but in all cases those differences were so small that I would never be able to successfully A/B them with a week or two gap in between. I accept that after a sufficiently-extensive CL2 listening period, there's a possibility that one's brain may adjust to the sound. But even if that were the case, would you really want that? Your brain and ears are already tuned to the natural sounds of everyday life. A good pair of headphones should be tuned to you, not the other way around.

Even using foam tips and accounting for the ~9 dB peak shift from a diffuse-field compensation, the rise in SPL between 2-5 kHz is still about 7 dB north of neutral. There is also a substantial drop into the lower treble (RHA's own measurements show a 30+ dB delta from 4 kHz ->10 kHz, although, as mentioned above, I suspect things won't actually be this bad for most listeners).

Varying the insertion depth will shift the location of the resonance peaks somewhat to the left or right. As shown above, ear-tips will alter the roll-off in the treble, and marginally alter the amplitude of the mid-range. Adding some significant resistance to the cable might drop the mid-range bump by another tiny margin, but IMHO, none of these effects are going to make an appreciable dent in a mid-range bump this large. While some level of mid-range boost is a feature of most IEMs (for the reasons discussed above), the level of this bump in the CL2 is an outlier with respect to any other headphone I own. At 4 kHz, it is more than 5 dB larger than the bump in the ER4XR, which is a deep-insertion IEM and so, arguably, has more reason for such a boost (the CL2 cannot insert as deeply into the ear canal as the Etymotic IEMs):



To some degree, I don't think there's necessarily a right or wrong with FR. Some people may love this CL2 tuning. But it would be interesting to know if a majority of people would prefer this large of a mid-range boost. This isn't an easy thing to test, because even a parametric EQ is a fairly crude tool without knowing exactly where those peaks fall, given one's particular eartips, ear canals, hearing sensitivities, etc. But I tested a couple of curves on the parametric EQ in UAPP on my V30, for example:



I found I needed to drop that mid-range bump by around 8 dB to get something that sounded roughly normal to my ears. After quickly toggling this EQ on and off... well, I can't imaging anybody would prefer the non-EQ'd version. But let's find out... You can vote here:

https://www.head-fi.org/threads/rha-cl2-tuning-preferences.894595/

BTW, please don't vote unless you've actually had a decent amount of listening time with the CL2, and have actually tried to EQ out some of that mid-range bump.

THD

Measurement of total harmonic distortion is not entirely straightforward, because THD varies with both frequency and amplitude (it generally gets worse as you crank up the volume, causing greater movement of the driver). For this reason, it's not really helpful to put too much significance on any single number. But given the level of effort needed to make multiple THD measurements at different amplitudes and considering how much I was paid to write this review ($0.00), I measured THD at only one SPL, which is near the higher-end of my usual listening volume. These are flat (Z-weighted) THD measurements for the CL2 at 80 dB:



These are very good - the CL2 has extremely low levels of THD. (Lower even than those of my all-time favorite headphones - the Shure KSE1500 electrostats. However, bear in mind that the KSE1500 has a very quiet amp and in practice, you might also have audible N from the DAC/amp you're using to drive your CL2.) I did detect a slight noise floor from the CL2 using my QP1R, but I'm insanely sensitive to hiss - I suspect most people wouldn't be bothered by this.

Impulse Response

I anticipated that a planar magnetic driver would exhibit a faster response than a balanced-armature, and certainly faster than a more traditional dynamic driver, (if Xelento's Tesla driver can be considered traditional?) but that's not the case here. The CL2 driver doesn't seem to have a lot of damping and exhibits quite a delayed response with some notable ringing, suggesting a larger/heavier/slower-moving diaphragm:






If you think you can't differentiate transients down to such a small number of microseconds, check out the following (there's a readme in amongst the test tracks):

https://www.dropbox.com/sh/qmkr6429gqpxfhe/AABxRdfmfgkOxoyTdYBsqZmba?dl=0

You might be surprised at how low you can go. You should certainly be able to discriminate lower than 1/(2*your_hearing_threshold_in_Hz).


Impedance

The CL2 is nominally rated at 15 Ohm, however, they have a relatively low sensitivity (89 dB/mW), so you will need a reasonable amount of juice to power them with a comfortable amount of headroom. Their impedance curve is impressively boring, i.e., almost completely flat apart from a tiny bump at 4 kHz, but its deviation is nowhere larger than 1 Ohm:



I had no problems driving the CL2 from my smartphone, but then again, my smartphone is an LG V30 For those that know the V30, the CL2's ~15 Ohm impedance triggers normal-device mode, which for me, was plenty loud enough, but there are options to trigger aux and even high-impedance mode, should you want even more power (https://www.head-fi.org/threads/music-apps-tips-and-tricks-for-the-lg-v30.868978/).




Summary

The CL2 is a beautiful-looking IEM that ticks almost every box you could think of in an IEM wish-list - with one big exception. It has a weird, mid-centric-boosted tuning that you can't do much about. The CL2 has very low levels of THD, but its FR just overwhelms everything else about the listening experience and this is something people might either love or hate. I'll just be diplomatic here and say that I didn't love its FR, which was really disappointing for me considering how good this IEM is for ergonomics, comfort, fit, isolation, etc. EQ helps, but isn't a panacea when most of your listening kit doesn't have a parametric EQ. For the price of the CL2, one shouldn't need to resort to EQ. I have a lot of respect for RHA's design team though and I see a lot of promise for the CL3. I would be all over that in a heartbeat if it were to give us a more conventional (flat, neutral or gently U/V-shaped) frequency response, or at least an option to tune it more in that direction.

A good initial trailblazing effort from RHA. I will certainly be first in the queue to check out the CL3...

The main purpose of this review is to provide a comparison to the FLC8S and answer the question: "Is the FLC8N worth upgrading to?". As an FLC8S owner, I have been searching the forums for months and failing to get an answer to that question. So I decided to answer it myself and document my findings in the process. I won't reproduce more glamorous photos of the FLC8N in compromising positions and various states of undress, as they already exist in the excellent review from @moonstar. Moonstar's review would be a better place to start for somebody who is unfamiliar with the FLC headphones.

Preamble

All my reviews are based on items purchased myself. I have never received payment or free merchandise in return for an "unbiased" review. There are two important reasons for this:

1. Personal integrity. We're all victims of subconscious bias. Of course you don't think that this affects you - that's because it's subconscious. But no matter how much a reviewer claims that the free pair of headphones in no way affected his/her opinion, you can be sure that it did. In addition, maintaining a steady stream of free merchandise necessitates that a reviewer simply can't be too negative about the products they review, even if they feel the product is a dud.
2. Nobody has ever offered me any free products to review. You can be sure that if they did, item 1), along with all my personal integrity, would instantly go out the window

The FLC8N

The FLC8N is a successor to the wonderful FLC8S triple-driver (two BA + one DD) IEM, which remains one of the most impressive value-for-money audio purchases you can make for less than $1000 (USD). So how do you follow-up one of the best-sounding and best-engineered headphones? From the plastic-housing of the 8S, FLC have now moved to a metal-housing for the 8N. The 8N is a similar-looking (but equally light and tiny) IEM that is also a triple driver (two BA + one DD) with the same unique three-part tuning system that allows for separate user-adjustment of the ultra-low frequencies (ULF), low frequencies (LF) and mids/highs. It does contain one new element, which is the presence of three little vent ports:



For those with limited patience or attention spans, here's the first burning question answered... Is the FLC8N worth upgrading to, if you already own the FLC8S?


The second burning question...


A Word about Measurements

I tested 2.5 mm balanced versions of both the FLC8S and FLC8N and measurements were made via a 2.5 mm to 3.5 mm adapter from a Hugo 2 operating as an external DAC to an iMac, with an external StarTech USB sound card and a Vibro Veritas coupler driving the REW software. All FR measurements are diffuse-field compensated using curves borrowed from InnerFidelity. The measurements were all made with SpinFit Cp100 eartips inserted to the same depth in the coupler and secured with putty before each reading.

IEMs are the one type of headphone where one can strictly reproduce the seal and insertion depth each time and this generally results in very good consistency between different measurements. The reader should note, however, that the measurements presented here won't exactly match the frequency response you'd experience in your own ear canal(s), which will be of a different shape and depth. However, the point here is to show the trends or differences from one headphone (or tuning system) to another. In that regard, these results should be easily repeatable by other measurement rigs.


FLC8N - Initial Impressions

In my opinion, FLC and its engineers (led by Forrest Wei) are geniuses in the field of IEM design. They created a wonderful-sounding, extensively-tunable IEM in the FLC8S that hasn't been rivaled at any price anywhere since. Well, until now However, the FLC team clearly lack marketing skills (at least in the English language) and seem to put out a rather weak message about their new products. For months now, people have been asking for information on the FLC8D, FLC8N and Celeste, and it's been like getting blood out of a stone. In particular, some important questions seem to have been ducked over and over. For example: 1) Will the FLC8N sound better than the FLC8S? 2) If so, how/why? The only answers that I'm aware of from FLC are 1) FLC8N has a wider soundstage and 2) this is because FLC8N uses a "newer" armature. When asked what was the improvement in the armature design, FLC didn't know, because it was proprietary to the armature manufacturer. So, all we're really expecting is apparently a wider soundstage(?).


The FLC8N earbuds are slightly more bulbous than the FLC8S, and the nozzles are about 1 mm shorter. That's probably not an issue for most people, but those of you with longer ear canals, bear in mind that you might lose some insertion depth. Personally, I love the ergonomics of the FLC8N and had no problems achieving an equivalent seal/insertion as with the FLC8S. The cable is also an improvement over that on the FLC8S, being slightly less springy. The pins are now a more standard 0.78 mm size, but the plug housing is a thicker, non-conventional shape which makes it unlikely that you'll easily be able to find (or solder your own) replacement cables and have them connect without looking a bit weird at the connection point.

Right out of the box, I put my favorite FLC8S filters onto the FLC8N. These were red ultra-low filter (ULF), modified black low filter (LF) and gunmetal mid/high filter. (The modified black was simply some acoustic damping foam inside the black LF in order to split the difference between the black - which I found too boomy - and the gray, which I found slightly lacking in sub-bass.) I will refer to these red, modified-black, gunmetal filters as R-MB-Gunmetal. In the measurements, you'll also see R-G-Gunmetal (red ULF, gray LF, gunmetal mid/high), R-G-Green (hopefully the nomenclature is obvious now?), R-G-Blue and R-G-Gold.

Ok, so with the R-MB-Gunmetal filters swapped over from my FLC8S, I expected to hear basically the same sound as my FLC8S. But I didn't. The FLC8N is somewhat more v-shaped than the FLC8S. More bass (particularly mid-bass) and more treble (particularly lower-treble). Now, I'm a fan of (gentle) v-shaped sound signatures, but it's all too easy to end up with an M-shaped sound signature where you're only actually boosting the mid-bass and the 5 kHz - 9 kHz resonance peaks, with an early roll-off in the sub-bass and upper treble; to my ears, this makes any type of headphone sound a little cheap, with boomy bass and borderline hot/sibilant treble. The FLC8N, with the R-MB-G tuning gets awfully close on both points. The increase in the 100 Hz region doesn't appear to be entirely explained away by the effect of the vent ports, as we'll see later when comparing FLC8S vs FLC8N. However, they seem to create a bit of an extra kick in the mid-bass. The following measurements were FLC8N with R-G-Gunmetal tuning:



The problem with the placement of the three new vent ports on the side of the FLC8N is that for some people they'll be open; for others, with larger ear canals, these may be pushed right against your concha bowl and be effectively blocked. The old adage that everybody hears things differently is definitely going to apply here.

As I wasn't enjoying the elevated bass on the FLC8N with the R-MB-Gunmetal tuning, I switched to R-G-Gunmetal filters. This sorted out my issue with the low frequencies, which still had more kick to them than the FLC8S with R-MB-Gunmetal filters, but I could happily live with that. I like this part of the frequency response and I think this is the right choice for the FLC8D (the fixed-frequency ULF/LF model). That should make the FLC8D a more tempting purchase, because I doubt I'd ever switch away from red ULF and gray LF filters now. However, I have so far not found a way of taming the lower treble. The only filter that gives less treble than the gunmetal mid/high filter is the blue, which tends to exhibit a massive 8.5 kHz resonance peak and causes too-early a roll-off beyond that point. Here's the effect of the mid/high-frequency filters in the FLC8N:



I should note that the above were all measured on the right channel. The channel matching on the FLC8N isn't all that great. Here are the L and R buds with identical filters, FR normalized at 1 kHz:



Manufacturing variation is understandable. But I think FLC could perhaps have taken a bit more time and effort in terms of quality control to ensure that the left and right channels were within a reasonable tolerance. The L and R measurements are different enough on the FLC8N that they look like they could almost have come from two completely different headphones. (Channel-matching isn't all that great on the FLC8S either, so you're somewhat rolling the dice on a purchase of either FLC8S or FLC8N.)

As for the increase in sound-stage width, I simply couldn't hear it. I tried with a variety of tracks and genres, including some binaural DSD recordings I have from Locatelli. If the sole purpose were listening to the differences in the sound-stage width, I would certainly fail that particular A/B test. I did notice a slightly more focussed sound stage, but the effects were fairly minor.


Frequency Response

This became a bit challenging because of the observed differences between the L and R buds on the FLC8N, so most of the following measurements represent an average of the L and R responses. Here's the all-important comparison of FR between the closest tunings I could achieve - FLC8S (R-MB-Gunmetal) and FLC8N (R-G-Gunmetal):



With respect to the FLC8S (again, always with my reference FLC8S R-MB-Gunmetal), the FLC8N (R-G-Gunmetal) has both elevated bass and treble. I find the FLC8S close to my ideal sound signature, but if there were one thing I'd change, I'd have less amplitude in the lower treble (5 kHz - 9 kHz) region. Unfortunately, the FLC8N goes in the other direction here

There seems to be no significant improvement in total harmonic distortion with the new model (in fact, the correlation might go the other way):



There aren't any notable differences in the spectral decay via waterfall plot:





The impulse response shows perhaps(?) marginally faster recovery with the FLC8N, although at the expense of higher-amplitude ringing.




Final Thoughts

There's a definite possibility you'll notice the difference between the FLC8S and the FLC8N. Some of those differences appear to be intentional design choices, but some additional changes might be heard simply as a result poor manufacturing tolerances (in which case, hearing a subjective improvement is going require a lucky roll of the dice). On average, it appears that the frequency response of the FLC8N is a subtle change from that of the FLC8S, mainly comprising of a slightly increased v-shaped sound signature. I don't mind the extra bass, but I don't really want the additional brightness, especially in the lower treble. YMMV in that regard. Even though my FLC8S with R-MB-Gunmetal tuning is closest to that of the FLC8N with R-G-Gunmetal tuning, I still slightly prefer the overall sound of my FLC8S. I can hear little change in soundstage width, but precision and focus of the soundstage does seem marginally better with the FLC8N. There appears to be no significant improvement in total harmonic distortion and the spectral decays and impulse responses are very similar. The choice of FLC8S vs FLC8N mainly boils down to whether you prefer a more `reference' or neutral sound, or a marginally more exciting, slightly more exaggerated v-shaped sound.

If you're in the market for a new IEM, relative to everything else out there, the FLC8N is still a very good headphone. It's just that, unless you prefer the marginally-different sound signature, it's not a significant improvement from its less-expensive predecessor.

The latest X7 Mark II from FiiO represents a step-up from the (already great) X7 in almost every regard. I would like to sincerely thank FiiO for allowing me to be part of the North American tour group. The loaned unit I received was returned at the end of my allotted time with the device, so my review was not in any way influenced by FiiO.

Please note that to save me from carpal-tunnel syndrome I'm going to refer to the X7 Mark II as the X7ii throughout most of the rest of this review. I'll also be referencing the original X7, X5iii (third generation X5), X3ii (second generation X3) and X1ii (second generation X1).

What's in the box?

Briefly, here's what you get:



Out of the box, the X7ii comes already seated in a clear TPU case with a screen protector already applied. These are really nice touches. The TPU case is similar in design to that which came with the X5iii and, in my opinion, is a step-up from anything that was available previously for the first generation X7. An alternative black leather-effect case is also included in the box, but to me, this looks a little cheap; I much prefer the TPU case. (FiiO - any plans to sell these parts separately?) Aesthetically, FiiO has the X7ii just about right. I love the volume control on the side, and the fact that the power button, now placed on the top, isn't as easily pressed by mistake when you're reaching for controls on the other side of the device (play/pause and track forward/reverse). The screen is similar to that of the X7, but doesn't protrude out as far, giving a sleeker overall look. The X7ii is overall slightly shorter and thinner than the original X7, but amazingly, FiiO has designed the X7ii so that all the amplifier modules from the original X7 will fit and work perfectly with the X7ii. FiiO also include a small fillet of metal which sandwiches against the side of the amp to blend in with the ridged left-hand edge where the volume, play & track controls live.

Wonderfully and amazingly, all functions on an Android cable remote now work properly! (play/pause, track forward/backward, volume up/down, including holding down to rapidly raise/lower volume). Thanks FiiO

A big feature that many will appreciate is two microSD card slots, supporting up to 256 Gb each (and possibly higher, with the imminent release of SanDisk's new 400 Gb cards). These fit in little SIM-like trays on the right side of the device. These look sleek, but remember that you will now need a paperclip or tray-removal tool every time you want to swap out a microSD card.

Amps

The X7ii comes with a new AM3A amp with both single-ended (3.5 mm) and balanced (2.5 mm) outputs. Given that all existing X7 amp modules would work with the X7ii, it's perhaps a little strange that there is no option (as there was with the X7) to buy the unit alone, i.e., without an included amp module. I hope that option becomes available later. The AM3A amp sounds great. From what brief A/B tests I was able to do through the balanced output via hot-swapping with AM3 (from the original X7), I couldn't tell too much difference in sound, beyond the fact that the AM3A balanced output has a slightly higher noise floor than that of AM3 (both have very quiet noise floors via the single-ended output). I'm aware of headfonic's review where they claim the AM3A has a quieter noise floor for IEMs because of the reduced power, with respect to the original AM3 amp. I can only tell you that is not the case for the unit I got, at least when listening via my SE846. You do need to be prepared for a little bit of a bump in the noise floor with the AM3A amp - though it's not as bad as that of the (otherwise great-sounding) QP1R. The strange circuit noise I occasionally noted on the X5iii single-ended output is thankfully gone on the X7ii.

Software

It would be fair to say that FiiO's X7 came with some software bugs and glitches. The original X7 was their first Android player and the software was continuing to evolve at and beyond launch. As a result, there were some notable issues initially, such as a failure to scan large music libraries, strange characters in file names and a period of time where gapless playback wasn't available in the more stable Android 5 system. Since then, FiiO has launched the X5iii and now the X7ii which have all been able to lean heavily on these earlier software developments and improvements, to the point that the X7ii feels much more stable right out of the box. That's not to say it is completely free of issues though...

Gapless playback isn't on by default, but needs to be turned on in the FiiO music app. It didn't work for me at first - there was a notable click between tracks, but this seems to have been fixed via a FW update to 1.0.0. At least, it's been working for me. (I'm aware that others are reporting the occasional click between tracks.)

The FiiO music player app is just awesome - it's easily my favorite Android music player. I'm still desperately hoping FiiO will one day release some version of this on the Google Play store, because I'd love to have it on my phone. That's not to say it doesn't occasionally crash and force you to restart it, but for the most part the odd crash has been tolerable.

Sound output via the FiiO music app takes a second to start on playback, or on scrubbing forwards/backwards in a track. Most of the time the pause is just silence, but sometimes you get a blast of high-pitched electronic noise. When scrubbing forwards/backwards in a track, that first second before playback starts at the designated track position is actually the tracking continuing at the last position when play left off (prior to scrubbing). In both cases, it's as if the X7ii is flushing its old playback buffer.

The idle power-off timer doesn't currently work at all. If you go past the allotted time, the player doesn't power down until you go to power it up again by hitting the power button (not having realized it never powered off in the first place). This can make the (relatively slow) start up just that bit more painful. I disabled the idle power-off until (hopefully) this feature can get fixed via a future firmware.

Fiber-optical Toslink output is a great new feature in the X7ii and it has worked perfectly for me, up to 96/24 PCM. Beyond that, rather than reporting it cannot output higher sample-rate PCM (or DSD), you just get weird garbled artifacts or digital noise from your external DAC/amp. (A graceful exit and error message would be preferable?)

WiFi is much stronger than in the X7, with support for 5 GHz networks, which really makes a difference in transfer speed. Unlike the X7, the X7ii appears to have zero RFI or EMI noise and no interference from nearby cellular radios either - great job FiiO One small caveat - I was unable to get the X7ii to connect to my smartphone's mobile hotspot when using WPA2-PSK encryption. It works with no WiFi encryption, but that's probably not a smart idea these days. I reported this issue to FiiO and they've said they're looking into it. I have occasionally found it useful to run Tidal via my smartphone's internet connection when out and about, so I hope this can be resolved at some point.

Digital-out via USB is something that was supposed to work, but is a giant crapshoot right now. It doesn't sound all that good anyway, but it also seems to get punctuated with interference, dropouts, and then it fairly reliably freezes the entire player. After one such episode it took me an entire day to get the X7ii to start charging again. I thought I'd broken it I eventually got it charging again, but have not tried USB digital output since. Digital output via coax works just fine:


BTW, the above setup sounds amazing, but takes quite a while to hook up. But even the quick, convenient, portable setup below sounds really, really good:



Sound

Here are a couple of frequency-response measurements I made comparing X7, X7ii and QP1R using SE846 and FLC 8S IEMs:




The wild impedance swings with the SE846 always cause some surprises and you can see the FR is in both cases more dissimilar with the X7ii than either of the other two DAPs. I did most of my listening with the FLC 8S via the balanced output.

Even though I was a big fan of the X7 with AM3, I always felt that its resolution and clarity didn't quite match that of my QP1R. With the X7ii, I no longer have that problem. Relative to the X7+AM3, the biggest change in sound I hear with the X7ii+AM3A is an increase in clarity. Sub-bass slam is perhaps still slightly less than that of the QP1R, but not to any meaningful level that I'd feel like I was missing something (like I was with the boring-sounding DX200). There is also a slight intonation issue I can hear with the new X7ii DAC, but it's something I only hear in the occasional bass note, it's not a deal-breaker and it's something I suspect most people would never pick up. In trying to track down the reason for the improved resolution with the X7ii, I swapped the amps over to run the AM3A from the X7 and the AM3 from the X7ii. The X7ii+AM3 still sounded notably cleaner. Hot-swapping the amps on a single device didn't yield any obvious changes in sound - both amps sounded pretty transparent. The amps are also fairly close in output levels, with the X7+AM3 needing volume level 39 to reach 85 dB on my FLC 8S with my white noise test track, and X7ii+AM3A needing volume level 42 to reach the same SPL. In conclusion, the ESS9028PRO DAC in the X7ii seems to be the most important and significant cause of the sound upgrade. I think X7 owners looking to see a significant improvement in sound quality by simply purchasing an AM3A amp are going to be disappointed.



Summary

The FiiO X7ii is a fantastic product and great value for money. It competes with the best DAPs I've heard (at any price) and has a richer feature set than any other DAP I'm aware of.

In the past, I've been critical of certain FiiO products. For example, I raked the X3ii over the coals for what I felt at the time was an expensive way to add very little (if any) sound-quality improvement over that of a good modern smartphone. Just when it seemed that all these devices (pagers, mp3 players, phones, GPS navigators, etc.) were converging into one device (i.e., the smartphone), Apple went and did something completely lame, dumping the headphone jack, forcing the use of Bluetooth and/or proprietary digital ports that require licensing - and giving up any pretense about caring more about audio than they do about money. Prior to this point, devices like the X3ii offered increased storage, a bit more power, but - to my ears - an unconvincing jump in sound quality - all of which required you to carry and charge another device. The X7ii leaves all these issues in the dust, with a meaningful boost in sound quality, a huge boost in storage capacity AND the ability to reliably stream (without interference noise) via third-party Android applications through a desktop-quality DAC. This is, in my opinion, exactly the right move from FiiO. However, given the surprisingly-good sound quality from the FiiO X1ii, this creates a rather awkward middle ground for the X3ii (or upcoming X3iii?) and now the X5iii as well. The X1ii is small enough that it could replace a phone when you're running or doing the yard work and don't want to haul around an enormous phablet (which most smartphones are becoming these days). The X7ii, at the other extreme, can afford to be bigger and bulkier because of the massive range of features, inputs/outputs and boost in resolution. I loved the X5iii's DAC, design and form factor, but was less impressed with its amplifier. The fact is the sound quality of the X1ii and X7ii are so good, I see little use for those in-between devices, i.e., the X3 and/or X5 DAPs.

If you're considering an X7ii, I can strongly recommend pulling the trigger. If you can't afford an X7ii and are thinking about an X3(ii or iii) or X5iii instead, I strongly recommend saving up your pennies until you can afford an X7ii.

For existing X7 owners (like me), this could be tough for your wallet. While the X7 does most of what the X7ii does, including the important parts (such as streaming via Tidal), the small differences add up. The one big new feature I can't easily forget after my time with the X7ii is the bump in sound quality as a result of the new DAC. Even though I've been very happy with my X7, I suspect I'm going to have to buy myself an X7ii at some point in the near future...

Preamble

I would like to be able to start by thanking Beyerdynamic for the free pair of Xelentos. Unfortunately, I can't, because Beyerdynamic didn't give me a free pair of Xelentos, so I had to buy my own But the silver lining for anybody reading this review is that I truly wasn't influenced at all by payment or free review samples I truly found them good enough to spend my own money on.

Beyerdynamic's Xelento took me by surprise. I'd heard of it, but had no particular interest in listening to yet another single dynamic driver headphone. I didn't expect to be impressed, but decided to pull up a chair at the Beyerdynamic booth at the 2017 SoCal CanJam, and this headphone stole the show for me. It was the best all-round headphone I listened to (including a few over-ear cans) and I ended up purchasing a pair shortly afterwards. I own the Xelento Remote, but all my listening tests and measurements were made using a custom T2 silver litz cable.


Fit, Comfort and Isolation

The Xelento's have fairly short nozzles. I would personally have preferred the nozzles to be fractionally longer, but once I got use to them, and with some heat treatment to my cable, I was able to get a fairly consistent fit with my favorite CP100-M SpinFit tips. Because of the short nozzles, deep insertion isn't really possible with the Xelento. The advantage to that is they're incredibly comfortable; the disadvantage is they don't isolate particularly well. In fact, you can tick that box twice, because they're also ported. So while they sound amazing, don't expect them to rival the sound isolation you'd get from an Etymotic.

Detachable MMCX Cable with Remote

There are two great aspects to this. 1) Xelento uses (almost) standard MMCX connectors. 2) The remote that comes with the "Xelento remote" version, quite remarkably, works perfectly on both Android and iOS devices. I'm not sure how Beyerdynamic did this. I've owned several cables with iOS remotes, Android remotes and even a few that work with both, but those that are both Android and iOS compatible have always had a switch on them to toggle between the two systems. The Beyerdynamic remote works perfectly on both operating systems without any need to toggle switches. I don't know if I'll ever use the iOS remote functionality (I'm no longer a fan of Apple, and if you care at all about audio, you shouldn't be either: https://www.head-fi.org/f/threads/apple-ditching-3-5mm-jack.795656/page-2#post-12708687), but it's a nice touch from Beyerdynamic to give all users that capability.

Now, about those MMCX connectors. I've never been a particular fan of these swiveling connectors. Any current or past owners of the awful (and awfully-overpriced) UE900s, will know what I mean. The MMCX connections on the UE900(s) were prone to drop-outs, no matter how much Deoxit you used. I'm sad to say, I experienced the same problem with the Xelento when attached to my various single-ended and balanced litz cables However, thankfully, there is an easy solution to this. The problem is caused by the center pin on the cable plug not extending quite far enough into the female socket on the Xelento. You'll know if this has happened, because the plug will swivel easily and loosely in the socket and/or you'll hear those drop-outs. What you need to do is take some very fine grit sandpaper and carefully (making sure to keep the sandpaper perfectly perpendicular to the axis of the plug), sand off a fraction of a millimeter from the plug (do not do anything to the Xelento earbud itself!). You want to shave off just a fraction of the plug's lip, shown in red below:



Be sure to clean off any debris and burrs (a can of compressed air will work) from the plug before you connect it to the socket on the Xelento. You should find you now get a snugger, tighter fit with less swivel. No more drop-outs and no more messing about with Deoxit

Sound

You really have to hear them for yourself! The Xelentos have a pretty exciting, gently V-shaped sound which stood out above everything else I heard at the SoCal CanJam. I'll go over this in more detail with comparisons to other competing products below.


Burn-In

I was advised by the Beyerdnamic rep to burn these in for about 80 hours prior to use. Burn-in with 80 hours of pink noise did make a measurable difference, however, it was pretty small - small enough that there'd be no way I could ever properly A/B those differences either side of an 80-hour gap. Burn-in seemed to increase the response at the frequency extremes for both left and right buds, but I don't want to make a big deal of this. The effects are small enough that I wouldn't worry about missing out on anything if you choose to use them straight out of the box.




Channel-Matching

Dang this is close. I've never seen a headphone that had such good matching between L and R buds:


Comparisons with Other IEMs























Summary

I found the Xelento to be easily one of the best-sounding IEMs you can currently buy - at any price. I do prefer Shure's KSE1500 for detail retrieval, but the Xelento isn't all that far behind. The only other non-electrostatic IEM I've heard that had this level of fit and sound quality is the outstanding FLC-8S. However, the FLC-8S comes with a caveat - you need to be prepared to spend quite a bit of time tuning it, and it has tiny filter components that can be easily lost. The FLC-8S isn't for the casual listener or for the faint of heart. The Xelento sounds fantastic right out of the box, and even though it can't be tuned, its frequency response is already very close to that of my ideal sound signature.

If you've never heard the Xelento, I highly recommend you try to get a demo with these outstanding headphones. I doubt you'd be disappointed.

Introduction 

 
The following impressions and measurements were made using a loaned X5iii unit which was returned to FiiO after the review. I want to say a huge thank you to FiiO for their bravery in selecting me for the North American review tour group 4 for the X5iii. I've never been very good at sugar-coating things, and I don't hold back when reviewing a product I don't like; I don't court controversy, but gushing exuberantly over every product you review doesn't paint an honest picture for a potential buyer. I've written some fairly scathing Amazon reviews of some earlier FiiO products, e.g., EX1 (nice sounding headphones, but I hated their ergonomics) and the X3 second gen (which I felt was overpriced, considering its sound wasn't all that much better - and its noise floor was worse - than my smartphone at the time). FiiO has taken some big steps since the X3ii, but without knowing what headphones are going to be used with the device, there still has to be a question mark over how much (if any) improvement in sound you're going to get with any dedicated DAP. For the right kind of headphone (explained below), the third-generation X5 could be your current best-choice in digital audio players. In order to save some typing, from this point on I'm going to refer to the X5 third generation as the X5iii. Please pay careful attention to the number of "i"s, because I'm also going to occasionally reference the X5ii
 
 
 
 

Hardware Design:

 
The hardware design of the X5iii is its strongest feature. It's a big change from the X5ii (second generation), with a move to a large touch-screen replacing the previous small screen and scroll wheel. Compared to its big brother (the X7), the X5iii is smaller, lighter, and fits easier in the hand. The volume control wheel is an awesome addition and is perfectly positioned to operate with your thumb when holding the player in your left hand. I've seen some people complain about the position of the hardware buttons and the risk of inadvertently pressing them, but I didn't have that problem. The only thing I'd wish for with the forward and back buttons is a long-press feature for forward/reverse scrubbing within a track. The two microSD card slots are a huge bonus, though there is a trade-off with the slick, streamlined design of the microSD card trays and the convenience of being able to just push an SD card out without need to find the right-sized paper-clip (a problem that I did encounter during an X5iii listening test with one of my headfi buddies). The X5iii screen has a slightly darker hue than that of the X7, but it still looks great and I didn't have any problem reading the screen, even in strong sunlight. Overall, I loved the design.
 

Software:

 
The software design of the X5iii follows closely from the developments of the Android 5 system used on the X7 and will feel very familiar to anybody that has used the X7. Unlike some vendors of far more expensive devices such as A&K and iBasso, FiiO offers a full Android 5 experience on the X5iii, which includes the Google Play store. The software highlight for me is FiiO's own music app. I understand that some people have issues with it, but for me, this is the best music app I have used on any device - ever. I absolutely love the FiiO music app and wish FiiO would offer this more generally via the Google Play Store. I would happily pay to have it on my Android smartphone. There are popular music players for Android (e.g., Neutron) that have far more features, but I very rarely - if ever - use those features, and in my experience these feature-rich players come with all sorts of bugs and issues. I love the simple, minimalist design of the FiiO music app; I love the option to automatically download synchronized lyrics, the integration with the Viper sound-effects module and the ability to rapidly change the volume by quickly swiping up/down on the right of the touch-screen. People may tell you that the software (mainly the FiiO music app) is buggy and that's probably still true - no DAP I've used is perfect in that regard - but what I find ironic about the complaints on the FiiO forums is that many of these issues are related to features that don't even exist in most other players. FiiO have worked hard to fix the issues in the Android 5 platform (and the FiiO music app) and a few of us on headfi have been helping to maintain a user-driven prioritized log of existing bugs on the X7. Those of you considering an X5iii might want to check out that list, because the vast majority of those issues will apply to the X5iii too. You will notice that most of the remaining bugs are not critical: http://www.head-fi.org/t/824174/known-bugs-issues-with-latest-x7-firmware
 

Listening/Measuring Tests:

 
I tested the X5iii using the following headphones:
 
 
I compared DAPs on the balanced-output IEMs, SPL-matched using REW and a Vibro Veritas coupler, by connecting the sources via a 4-pole triple-throw switch:
 

 
Each IEM tested with BAL-out was also tested on SE-out using its cable adapter:
 

 
Line-out into the KSE1500 was sent through a 3-pole triple-throw switch:
 

 

DAPs Compared:

 
 
I haven't listed MSRPs for the above DAPs, but some here are much more expensive than the X5iii and perhaps that might seem unfair. However, most potential buyers are still going to want to know how the X5iii compares in term of sound with the best currently available DAPs. It's a credit to the X5iii that with many headphones (there's a caveat, explained later), the X5iii compares remarkably well.
 
Of all the DAPs tested, the X5iii was the slowest to boot-up. It also takes some time to scan your music library (for my library of ~ 200 Gb worth of mostly FLAC files it took a bit less than 5 minutes, compared with ~ 1 minute on the DX200). However, music scanning is something you'll typically only do once (incremental updates are quicker) and, once booted, the X5iii has the fastest, most responsive UI of any DAP I've ever used.
 
 
 
 

Listening Tests:

 
Most of my critical listening was done with my IEMs using my four-pole/double-throw switch to alternate a balanced output from two DAPs into a single set of headphones. The output was SPL-matched as close as possible (typically only within about 0.2 dB because of the discrete volume steps in all of these DAPs). I did almost all of my listening via balanced-out. I need to confess that I did not do a carefully controlled test of SE vs BAL on the X5iii. I’d really have needed two X5iii units to test these, properly SPL-matched. I just don’t trust my own brain when switching from two devices with a long pause while I have to adjust the gain. In all honesty, from the very brief and unscientific tests I did, I couldn’t detect much difference between BAL-out and SE-out - with one exception… When using sensitive IEMs, there was a noticeable noise floor on the BAL (2.5 mm) output. I’ve come to expect that with my SE846, but this was particularly noticeable, even with my FLC 8S. The noise floor with the SE846 was loud enough that I could even hear it over quiet passages of music. That's the only major issue I had with the X5iii . It has an outstanding design, DAC and feature set, but it really deserves a better amp.
 
I also noticed something odd with the SE (3.5 mm) output, which was some circuit noise that was apparent even when all radios were off and no other transmitting devices were within close proximity. This circuit noise was quite distinct from the typical white noise one hears as a noise-floor from an amp. The SE-out’s noise floor was less than that of the BAL-out, but its circuit noise contained some unpleasant tonal components, with a peak around 1.6 kHz. This circuit noise was present on the SE-out, independent of the volume setting, but wasn't present at all from the BAL-out. I’ve not seen anybody else report this, so it’s possible this was a fault with the North American review tour group 4 unit. (This was an issue I reported to FiiO, but something I never got any resolution on.) In any case, unless you have sensitive IEMs, it likely wouldn’t be a problem: it was audible on the SE846, faint on the FLC 8S and inaudible on the ER4XR.
 
I did do some listening via Tidal and, as with the X7, there can also be some WiFi interference noise with more sensitive IEMs. It rarely bothered me, as it’s only really apparent before playback starts and can always be eliminated completely by downloading the tracks you want and then switching to offline playback. With regular music playback, the biggest sonic differences between the X5iii were also apparent with the SE846, most likely due to the fact that the SE846 has an extremely low impedance (~ 7 Ohm @ 1kHz) and also wild impedance swings with frequency. The X5iii sounds slightly warmer than the X7+am3 or DX200 - so, taking the edge off its treble and/or boosting its bass, depending on how you want to look at it. This is a likely result of the X5iii’s slightly higher output impedance, but it really isn’t a problem for me. The differences are pretty small and could be a positive or a negative, depending on your sound preferences. These sonic differences show up in measurements too:
 

 
The sonic differences were extremely minor with the FLC 8S, with noise floor (BAL-out) and circuit noise (SE-out) also reduced in comparison to that from the SE846. With the ER4XR, I couldn't reliably A/B any differences at all (and noise floor and circuit noise were totally inaudible). Could you hear the differences with the ER4XR? If they’re there at these volume levels, they’re very, very subtle:
 

 
The X5iii’s digital output via USB doesn't appear to be working yet (perhaps this will get addressed in a future firmware?) so I wasn't able to test USB DAC mode feeding the KSE1500, but I did do some listening tests using the X5iii’s LO into the KSE1500. Wow. I was very pleasantly surprised by the X5iii’s DAC. To my ears, the X5iii’s LO beat that of the DX200, with a more enjoyable sound and better extension at both ends of the frequency spectrum. (The KSE1500 always seems to have an extremely clean, black background, no matter the source.) Given its small size and excellent DAC, the X5iii would be a great choice for someone who already owns a good amp (such as the KSE1500’s electrostatic amp).
 

Summary:

 
The X5iii is a fantastic product, but it does have a slight weakness, which is its amp. Having said that, the weakness in the X5iii’s amp probably isn't going to be relevant, or even perceptible, to most people. Listeners with sensitive IEMs might want to look elsewhere, and those with harder-to-drive headphones might want to take it for a test drive first. But for a very large range of headphones in between, the X5iii has to be one of the most feature-packed and best value-for-money DAPs on the market right now.