Fit and Comfort

Seen above is the right channel of the 64 Audio U18t installed into an artificial ear mold, and I am using the size M black silicone ear tips included with the IEMs here. This is my typical combination for personal use too, and illustrates the nature of the achievable fit with these IEMs. I do have average-sized ears, and the ear mold above represents my own experiences well enough as a proxy. The semi-custom design of the shells shaped for the ear concha with ergonomics in mind works out well for most ear types, especially given the smaller size of the shells compared to most IEMs on the market. In general, the soft curves and design make for at least two to three points of contact with the ear, which adds support and keeps the IEMs in place. The longer nozzle also has the the ear tips go in far enough to make for a better fit and seal, although the company's modular Apex technology will ultimately decide how good the seal turns out to be. Even with the m20 Apex module, I would not describe these as any more than semi-open IEMs, so passive isolation isn't the highest by any means. However, this also makes the U18t extremely comfortable over even long listening sessions, as any built-up air pressure in the ear canals is released in a controlled fashion. As such, there is no sense of physical fatigue with a good fit. The angled connectors on the cable also work very well, as does the pre-formed memory-wire formation, which has the cable naturally go over and behind the ears as seen above. Pair this with the cable clip and cinch, and you now have a securely installed set of IEMs you may not want to remove, especially given these barely come to ~7 g each.

Audio Performance

Audio Hardware

64 Audio has an older YouTube video that briefly shows the making of the U18t, but what you need to take away from it is that these are completely hand-assembled. The U18t was also the world's first 18-driver IEM at launch, and continues to be among the very highest in driver count today. As per the name, there are 18 balanced armature (BA) drivers per side, with 64 Audio going with a set of eight BA drivers for the low frequencies, another set of eight for the mids, one for the transition from the mids to the high, and a tia BA for the high frequencies. Aside from Apex, tia is the biggest technology present in the U18t. It stands for Tubeless In-ear Audio and comprises three components, which can be seen in the image above—there is the actual driver inside a specifically designed chamber and the single-bore fitted into the nozzle. This setup aims to eliminate unwanted resonance effects BA drivers have inside tube enclosures, and has been put to good use in other 64 Audio IEMs. Missing here but present in newer 64 Audio IEMs is the Linear Impedance Design (LID), a proprietary circuit that corrects any driver-specific non-linear impedance behavior, which is especially relevant in IEMs employing a large number of BA drivers as it makes the output impedance of the source a big factor in the overall sound signature.

Driving the hardware is extremely easy even compared to most IEMs, with a rated impedance of just 9 Ω (at 1 kHz) and sensitivity of 111 dB/mW, again at 1 kHz, presumably measured at the drum reference point. In practice, a dedicated source isn't about power output with these, but rather getting one with low enough output impedance. That said, the absence of LID combined with these easily-driven drivers makes for a system you can customize the sound signature of, especially in the low frequencies, based on the source, any in-line resistor, and even cable contacts. Any halfway decent portable DAC/amp will suit your needs perfectly for when you find yourself without an available 3.5 mm audio jack on the go. A better source compared to the line out from a laptop or PC, or even a DAP, may worth considering. If not using them on the go, the shorter cable included with IEMs might be a potential handicap when connecting to a PC as the audio source, and a cable upgrade may be called for even assuming your PC has a decent DAC/amplifier setup.

Frequency Measurement and Listening

I will mention that I have a general preference for a warm-neutral signature emphasizing a slightly elevated bass and smooth treble range with detailed mids and good tonal separation. I also generally prefer instrumental music over vocals, with favored genres including jazz and classical music.


Our reproducible testing methodology begins with a calibrated IEC711 audio coupler/artificial ear IEMs can feed into enough for decent isolation. The audio coupler feeds into a USB soundcard (or FiiO K3), which in turn goes to a laptop that has ARTA and REW running and the earphones connected to the laptop through the soundcard (or FiiO K3). I begin with an impulse measurement to test for signal fidelity, calibrate the sound card and channel output, account for floor noise, and finally test the frequency response of each channel separately. Octave smoothing is at the 1/12th setting, which nets a good balance of detail and noise not being identified as useful data. Also, the default tuning was used for testing, and no app-based settings were chosen unless specifically mentioned. Each sample of interest is tested thrice with separate mounts to account for any fit issues, and an average is taken of the three individual measurements for statistical accuracy. For IEMs, I am also using the appropriate ear mold fitted to the audio coupler for a separate test to compare how the IEMs fare when installed in a pinna geometry and not just the audio coupler. The raw data is then exported from REW and plotted in OriginPro for easier comparison.


The careful reader of my IEM reviews would have noticed the inclusion of the FiiO K3 portable DAC/amp as an optional source above. I knew going in that the absence of LID and the high BA driver count in the 64 Audio U18t would challenge my soundcard's output impedance. Listening to these off the Qudelix-5K and FiiO K3 confirmed as much since there was no way the frequency response off the soundcard should have dropped as much in the bass and sub-bass. This is a great example of not relying on frequency measurements to frame your thoughts as a reviewer by always using your own listening experience as the primary source. As such, switching to a lower output impedance source solved the issue of frequency response measurements that matched my listening experiences fully, and note that this shift in the two measurements was reproducible across different Apex modules, as well as on the right channel. All subsequent measurements shown are with the FiiO K3.


The IEC711 is such that you can't really compare these results with most other test setups, especially those using a head and torso simulator (HATS). The raw dB numbers are also quite contingent on the set volume, gain levels, and sensitivity of the system. What is more useful information is how the left and right channels work across the rated frequency response in the 64 Audio U18t, or at least the useful part. The left channel was separately tested from the right one, and colored differently for contrast. I did my best to ensure an identical fit for both inside the IEC711 orifice, so note how the two channels are identical pretty much across the entire range. In fact, it's only well past the coupler resonance, where things are to be taken with a grain of salt as it is, that a small measurable difference appears. In practice, I could not tell the difference even when actively looking for it. Given the hand-assembled nature of the U18t, it is no stretch to presume the drivers are hand-matched in this flagship set. This was also the case with the other sample from before, so channel matching is topnotch here. Measurements taken after 40 hours of testing, including with these playing a mix of various songs as well as white or pink noise and sine sweeps, showed no difference. There was no perceived burn-in effect thus, and none was measurable, either. The response with the artificial pinna in place matches the ideal scenario in the coupler very well too, including with the typically seen resonance shift minimized to where it's an indicator of how good the seal was when installed in the pinna simulator.


This is the average frequency response for both channels of the 64 Audio U18t plotted against my personal target taken from VSG.squig.link, which also gives you an idea of my personal preferences to better correlate any possible biases. The tuning of a set of headphones or earphones does not have to match my target as long as it is tuned with some direction, makes sense, and is executed well. After all, no one set will appeal to everyone, and having different options is what makes this hobby so interesting and hard to quantify. With the U18t, things are all the more complex because Apex modules are involved. As such, I have four different graphs, which can all be examined separately on a per-channel basis for the U18t's frequency response with the m20, m15, or mX modules installed, as well as the control case with no Apex module at all.

The U18t is intended to be one of the most accurate monitors for music, making the "M" in "IEM" a legitimate claim here. This typically results in a neutral and faithful reproduction of the source music, in great detail, and does not necessarily mean it's a fun set to listen to. Indeed, other than detailed, I would describe the U18t as smooth, but even that has a caveat. Having the m15 Apex module pre-installed was the right way to go since it adds decent bass extension courtesy the 15 dB rated isolation and makes for the most balanced tuning of the lot. Of course, the control case without any module makes this an open IEM, so it naturally has the least bass presence. The other two, as the numbers suggest, follow the trend to where the m20 module provides the most bass response and the mX module the least. Unfortunately, I am not happy with either of these at all. The m20 Apex module makes the bass come off bloated compared to everything else, while the MX feels more analogous to an attempt at ambient modes in a TWS set than anything else. The m15 module wins mostly by process of elimination, but this is not to say that the Apex technology is not innovative; rather, the tuning of the U18t isn't a good base for making the most of it.

The U18t with m15 Apex modules is slightly warm-neutral thus, with a ~7 dB bass elevation from the lowest point in the mids. The bass quality is very good even before we consider that this is an all-BA set with the associated BA timbre. Notes decay fast, and there is resolving power to spare as you go about using the U18t for listening to the most complex of fusion music; say, different concurrently playing individual samples that need to be combined, as any changes are adapted to the master track. I can certainly see the set working great for music-monitoring and production, and it ends up plenty fun when listening to the likes of smoother music with bass guitars and kick drums alike. That may sound like an oxymoron, but try listening to some funk music and classic rock before you complain.

The transition to the mids is impeccable even with what is a massive set of drivers leading to another, and male vocals come off ever so slightly forward to add some presence when played alongside instruments. In fact, given the variable nature in the low frequencies, this is a set that makes the term "midhead" feel relevant. There is so much range for instruments here that I was left astounded by how well the likes of faster symphony music played with different instrument classes building upon each other and never clashing. The technical performance is also a masterclass of how to get it right, and one in how to achieve it with an all-BA set at that. Imaging was so accurate that the Audeze Euclid finally got bested after months of IEM reviews, and the soundstage also ends up being exceptionally large for the form factor of these IEMs. While not the widest set of IEMs I've heard, it is one of the tallest and deepest to where this also makes for a very good set for movie watching or playing games.

The tuning woes return in the upper mids, where the pinna gain compensation looks and feels lacking in many ways. Even as someone who appreciates a more pronounced 3 kHz peak for the likes of female vocals and jazz or eastern pop music, what 64 Audio provides out of the box will only just satisfy the other side of the spectrum, those preferring metal and/or rock music more. Thankfully, the U18t takes EQ like a champ, and this is where I used Equalizer Apo with a broad peak added. There are no complaints about the rest of the treble tuning, with tia showing itself to be another highly innovative technology that works exceptionally well. There have been no electret tweeter implementations that have provided the treble quality I desire out of IEMs, and then a single tube-less BA comes along and does it! I want to see a range with good treble extension, smooth peaks, and valleys that eventually drop with higher frequencies, and resolution that is separate from an elevated treble response, which many seem to confuse it for. Piano key fundamentals need to shine as higher-pitched vocal harmonics decay away, and then string instruments need to get some hang-time as well. This is an extremely challenging ask only the most competent headphones manage, but the 64 Audio U18t does it!


The spectrogram for the U18t with the m15 Apex module to an extent relates the decay and resolution, but keep in mind that this is directly taken from REW and the Y-axis is milliseconds (ms), with the top-down view of the frequency response decay. This is a fast set that is still smooth, similar to driving a V8 luxury car at 60 MPH without noticing any of the road noise or bumps. The engine here is clearly strong thus, but I wish the tuning was better. How's that for a fitting pun? I understand the newer U18s has tackled tuning head-on and implemented LID to account for a linear impedance across the entire frequency range.

Comparisons

The 64 Audio U18t is the most expensive set of IEMs I have on hand. As such, the natural comparison would be with other flagship-class IEMs I have on me as well. I am taking the liberty of including two sets that have not yet been reviewed: the Effect Audio Axiom for $1500, made in Singapore, and ThieAudio V16 Divinity for $1500, made in China, in addition to the recently covered Lime Ears Pneuma for ~$2000, made in Poland. Of the four, the Effect Audio Axiom is the biggest culprit by far, and I can't recommend buying it for the asking price because of the poor tuning and fatiguing listening experience. It does have one neat feature that will be discussed in its eventual review, but does not even come with a cable at that asking price! The Lime Ears Pneuma has a bass switch that changes the tonality in the lower frequencies similar to the U18t, but it doesn't come off as natural or fun to where some vocals can be nasal-sounding, and it has the worst treble range of the four here. It is also outperformed technically by the U18t, but that could be said of every other set here. The ThieAudio V16 Divinity is where things really step up, with this set having the best tuning of the lot and aiming for a mix of fun and accuracy alike. The U18t bests it in detail and even adds some air in the upper treble, so it comes down to pricing more than anything else. The V16 Divinity also comes with a modular cable and happens to be an all-BA set with a large number of drivers. As such, expect more comparisons and thoughts of the two in the V16 Divinity review.


Given the U18t is over five years old now and the IEM industry has gone through many an upheaval since, I sought to see how it hangs with the tribrids that have popped up since. I omitted the excellent but highly analytical ThieAudio Monarch for $730, made in China, since it's tuned very differently from the others here. The XENNS UP for $700, made in China, and LETSHUOER EJ07M for $600, made in China, are tuned more for audiophiles with a bass-boost and the use of electrostatic tweeters (electrets) for treble extension. The treble extension is where the 64 Audio tia puts them all to shame, especially when it comes to the XENNS UP and its inherent darkness in the treble. Both of those are more fun to listen to in the bass compared to the U18t, however, and come very close in speed, punch, and timbre, too. The Astell & Kern AK ZERO1 for $700, made in South Korea, is a fairly unique set using a micro-planar driver, dynamic driver, and two BA drivers, and more neutral-bright in tuning to primarily cater to the preferences of its home audience. It's an acquired taste tonally, even more so than the U18t, and does generally worse than the U18t in every single way aside from pricing.