This isn't really a formal review. It's just a fitting place to post my experience with the product so that I can link to it in forum discussions without having to retype the same comments over again. I suppose it kinda sorta works as a review.


It used to be $99?
The EO-IG955 is the earphone that originally came bundled with the Samsung Galaxy S8. It's also sold as a separate product, retailing for around 100 USD. Samsung acquired Harman and its subsidiaries, including AKG, who was responsible for tuning the EO-IG955's sound.


The EO-IG955 is also sold by itself without the box and the case. I got a three-pack from e-Bay for $20, but I've seen it sold for even less than that. Not bad for an IEM that usually retails for $99. But I hear that there are many fakes out there, and there are even guides online to help you spot one.

So what do we get in the box? The EO-IG955 has two transducers: one 11mm-diameter driver for the low frequencies, and one 8mm driver for the treble.


Zerodecibel has a nice tear-down on their site: link


A word or two about the measurements
The first two sets of measurements that I found were at zerodecibel: link and The Ear-fi Blog: link. They provide lots of useful information like impedance vs. frequency, and distortion. I don't think it's the complete story though, because they each measured only one sample, probably from a retail box. Here's what I got from the three-pack:


The bass response is obviously inconsistent, especially compared to other OEM-bundled IEMs like the LG Quadbeat 3 and the Sony MH755, whose performances are more repeatable. One of my samples has too much bass for my liking, and another one has mismatched channels. I'll call these samples 3 and 2, respectively.

Some of you are probably itching to blame Samsung's bad quality control. Are these dirt cheap IEMs the factory rejects that failed to pass QC? Or is the IEM's design to blame—that it's not robust, and that small changes in venting cause large changes in the response? I don't mind too much. My expectations weren't that high for such an inexpensive e-Bay 3-pack. Plus, you can sometimes fix the bass response. More on this later.

At least I got one pair that sounded fine from the start. Sample 1, the one with the least amount of bass, sounds best to me:


The EO-IG955 responds predictably to simple, reversible mods. Blocking the rear vent and partially or completely sealing the seam between the front and the rear of the IEM has this effect on the bass:


Here's where to make the seal:




Varying the leakage gives us some control over the bass response. Thus, we have a way of reducing the bass on sample 3 and fixing the mismatch on sample 2.

The EO-IG955 also responds nicely to dampers placed in front of the output tube. Use your favorite damping material: tea bag filters, microfiber cloth, micropore tape, etc. I'm lazy, so I just put a little bit of earbud foam inside the silicone sleeve, and placed as close to the grille as possible.


Some folks find the response at 3 kHz too intense, so here's a way to reduce it. The default response is a good match for my hearing, so don't do this. What about the spike at 8 kHz? That comes from the ear canal's half-wavelength resonance. A foam tip can damp its response:


It's getting close to the Harman IE target, so I should like this smoother response, right? Not so fast…


Insertion depth or the lack thereof
When I listened to sample 1, I thought it sounded fine. If I were being picky, I’d say that the treble sounds a bit off. Some sounds were emphasized, and some sounds were withdrawn. Vocals had a slightly scratchy character, as if the singers were straining. Snare drums sounded a bit too sharp with their attack, perhaps with too much snap, and they took on a hollow character, as if listening through a tube. At the same time cymbals and hi hats sounded suppressed. Shakers sounded somewhat withdrawn, and they took on the same scratchy and hollow sound. Very often, IEMs overdo the sparkle, sometimes annoyingly so. Here’s, it’s the reverse.

(Not that it matters much, but among the tracks that I used were Bird on a Wire and Cousin Dupree. I don’t necessarily listen to these artists, but Harman’s research had indicated that their listeners, novice and trained alike, were more discriminating when they evaluated headphones with these tracks, compared to others on their play list.)

After investigating with a sine sweep, I found that I hear the peak at 6.3 kHz instead of 8 kHz when I inserted it in my ears as deep as I could. The EO-IG955’s housing and strain relief make deeper insertion impossible. (To verify this, I used a program like Sinegen, or a finer version of this online tone generator: https://www.szynalski.com/tone-generator/). I later made a few more measurements, this time with shallow insertion. The following is the closest I got to what I feel I actually hear:


There's a big slide in the response after 7 kHz. I'm actually hearing the peak lower, so 7 kHz sounds a lot softer to me than 6 kHz. The smoothest response comes when the insertion depth causes an 8 kHz peak. That’s almost 6 mm deeper that what’s possible with my ears. Any deeper, and we get a dip in the response at 7 kHz. This can be seen in zerodecibel’s and The Ear-fi Blog’s measurements, which show the peak at 9k. The EO-IG955 is a shallow-insertion IEM. I doubt most adults can get it deep enough in their ears for the heard response to match those measurements.

The 6.3 kHz peak that I’m hearing doesn’t sound as severe as it looks in the graph. Using a parametric EQ to flatten it and smoothen the treble response makes a subtle difference, even though the adjustment is several dB. But the sine sweep sounds a lot smoother afterwards, and it makes all the difference. Subtle change as it is, the treble now sounds more natural, with the realism restored.

Unfortunately, a system-wide parametric EQ still isn’t available for my mobile devices unless it’s built into the earphones themselves, like on Jaybird IEMs. The Radsone ES100’s 10-band graphic EQ can’t make these fine adjustments. They’ve announced that a parametric EQ is planned for a future firmware update, but that was at least eight months ago; it’s still vaporware as of this writing.

I found that an absorbent foam tip brings the response closer to what I like, and makes it easier to fix with a graphic equalizer:


The TX400 damps the ear canal resonance, makes the transition less abrupt, and brings up the response from 8-10 kHz. Indeed, the sine sweep sounds smoother with the TX400. The response above 10k isn’t as strong, so there isn’t as much “air” as before. But overall, it sounds less wrong, and it’s easier to EQ.


Comparisons
The obvious comparison is with other OEM-bundled earphones, like the LG Quadbeat series and the Sony MH series. Here are some that come close to my sample 1:


The MH755 (my samples at least) have too much low bass. Fortunately, it’s easy to reduce with EQ or simple acoustic mods. The peak from the ear canal resonance is just a small bump unless you get a sample with more damping at the output tube's opening. It’s more forgiving of differences in insertion depth than the LG or the Samsung. I can insert it deeper in the ear than the EO-IG955, especially with the small tips, because the housing and the strain relief don’t get in the way too much. The high frequency extension is ok, but it’s not as strong as on the Samsung, with its dedicated tweeter.

The LG Quadbeat 3 sounds similar to the MH755, but its deep bass isn’t as crazy. The peak from the canal resonance is more exposed, but It’s not that noticeable unless you’re listening for it or your music has a lot of content at that frequency. My QB3 is the variant tuned by AKG. It has less bass than the non-AKG model, but it has a noticeable peak at 4.5 kHz. I’ve heard the regular QB3, which I prefer when I shelve down its bass.

I tossed a Moondrop Spaceship in the comparison even though it’s not an OEM-bundled earphone, because it’s cheap and its frequency response is similar from 100 Hz to 3 kHz. It sounds closer to my EO-IG955 sample 1 than the Moondrop Crescent. Its response in the bottom octave is the weakest here, and you can see the roll-off in the graph. Maybe a partial cover of its front vent will bring it more in line with the rest of the spectrum. I used to have another pair of the Spaceship, and its bass was similar. The treble, however, is different between samples—different, even between channels of the same pair. It’s not as consistent with channel matching as the Crescent, judging from my pair, and measurements I’ve seen online. Its treble is like the MH755’s, with the not-so-apparent peak from the canal resonance. Like the MH755, it’s also more forgiving of differences in insertion depth than the Samsung. The Spaceship isn’t as airy sounding as the Crescent, or the MH755 for that matter, let alone the Samsung.

In the graph, you can see in light gray that the EO-IG955’s bass can be very different from sample to sample. It also shows what can happen to the treble depending on how deeply it is inserted in the ear.


Wrap up
Overall, I think this is a good sounding IEM. The engineering needed for a two-way design seemed to have been done competently, which is more than I can say for some cheap multi-driver chi-fi designs. Best of all is the value: the IEM that comes in a $99 box set can be had dirt cheap online.

Update: August 17, 2019

Ok, it's sort of like a review. But not really. Technically it is?

I think it sounds good. I hear the canal resonance a little over 7 kHz, because I can’t insert it much deeper with the stock tips. I tried to match this with my measurement. The peak doesn’t sound as severe to me compared to what’s in the graph. You can determine where you’re hearing the resonance with a sine sweep. I created one that can be manually controlled: the link is in my sig. It also has a filter that can help determine the peak’s height in dB. Often, the center frequency isn’t exactly at the same spot for the left and right channels, and this can be heard as the sound moving from left to right (or the reverse) during the sweep.



Tips and tricks
The frequency response didn't change much with other silicone tips:






I like the sound better with deeper insertion, which I can get with the small Sony tips. The front vent seems to have been partially blocked by the silicone sleeve in the measurement with the small Sony. Here’s what it looks like when it’s blocked more:


The Sony tips aren’t an exact fit for the Crescent’s nozzle, so it’s a bit tight. I can avoid blocking the front vent by not fitting the sleeve all the way to the base of the barrel. This also helps with getting a deeper insertion in the ear to avoid frequencies where the resonance makes vocals too sibilant.




I tried it with Comly foam. TX400 is a tight fit over the barrel, so my guess is that it deforms the foam in a way that hurts the response above 10 kHz. The universal-fit Comply Sport Pro with Smart Core works better here:



The sound with the TX400 is way too dark. Comply Sport Pro makes it sound dark too, but not as much. The resulting response with the foam tips is a lot easier to EQ, compared to silicone. The spike from the ear canal resonance balances the elevated low end. When the foam tip flattens that peak, it sounds like you turned up the bass. But if a system-wide equalizer were available, this is how I would use the Moondrop Crescent. For me, the Comply Sport Pro is the least uncomfortable of the tips.



This result is a bit disappointing. Playing with front vents and poking holes through meshes have acceptable risk with dirt cheap IEMs. But for a $30 model that takes weeks to deliver, I think I’ll let someone else go first with the non-reversible mods.


Some comparisons
Here’s what the frequency response looks like with the closest-sounding IEMs I have:


I didn’t include the Sony MH1 in the comparison because my units don’t perform as designed and have to be fixed with mods, even though they behave like the typical samples that you can buy today. The Moondrop Crescent doesn’t sound as v-shaped as the others shown in the graph, especially the Philips. But the SHE8105 can be modded to sound close (see here: https://www.head-fi.org/threads/sou...odding-headphones.694963/page-4#post-14744462)


It seems that the Crescent is tuned closer to Usound than to the 2017 Harman IE target. I prefer it when the 8105 is modded to sound closer to Harman than Usound, but with less bass. It doesn’t have the Crescent’s air in the top octave, but there’s not as much musical content up there, and a lot of folks can’t hear them anymore. Instead of the Crescent’s resonant spike, the energy is more spread out across a wider frequency band. The 8105 mod has a smoother and livelier response up to 12k, where there is more to be heard. I haven’t decided whether I like the broad peak around 3k.

I'll update this report after I spend more time with the Moondrop Crescent. I've only been listening to it for two weeks.


Update, March 4:

I didn’t see a review of this old item or a suitable thread in the equipment forums, so I’m posting this here, mostly so I can have links to information for at least two conversations I’m having, without having to retype the same post. The content might be of interest to existing Jaybird X3 owners anyway.

This thing has been out for a while now, and it has been reviewed many, many times. For the most part, the reviews have been positive. The Jaybird X3 has been discontinued and replaced with the waterproof X4. This write-up is mostly for holding observations and measurements of the product that I haven’t found anywhere else online. It’s also a discussion of its built-in 5-band fully-parametric equalizer, which can save a preset on the X3 itself.


After over two years of on and off gym use, this unit has developed a channel imbalance in the area where human hearing is most sensitive. In practice, I really don’t notice it unless I’m paying attention and listening for it. The response is intentionally v-shaped. Maybe this is what people really want when they’re running or in the gym.


The shiny foam tips don’t do too good a job of damping the ear canal resonance when they’re new. The built-in parametric EQ can cancel the peak on one channel, but there’s no guarantee that the resonance in the other channel will be at the same center frequency, or even if it will be in the same spot the next time the IEM is worn.


The foam tips lose their shine and start to become more acoustically absorbent after some use. It changes the treble, for the better IMO. The resulting frequency response is smoother, and so is easier to EQ.


Comply sells the Sport Pro tips specifically for Jaybird IEMs. They smooth out the treble better than the bundled foam.


Comply also sells the generic Sport Pro tips with Smart Core, which can be used with all sorts of IEMs with different barrel diameters. The result is similar.

The parametric equalizer in the Jaybird MySound app is very flexible, but frustrating to use if you want to fine tune the response to your personal preference. The frequency markings on the horizontal axis aren’t too helpful because it’s very hard to tell where they are exactly. Worse yet, the vertical axis doesn’t even have any indication in dB how much the adjustment is.

Even with measurement gear, it can take a while to dial in precise settings. I played white noise from an iPhone with the X3 inside an IEC 60318-4-compliant coupler while measuring the response in Room EQ Wizard (REW). If you use the highest resolution settings in REW with lots of averaging, it takes too long for the EQ adjustment to settle on the screen. With fewer FFT points and fewer averages, the result appears on the display faster, but it’s fuzzy even with smoothing. Again, it’s hard to tell what the EQ adjustment really is. The FFTPlot iOS app is more responsive.

After a while I came up with these presets:


To approach the target, I cut down the bass, the response around 200 Hz, and the peak near 3 kHz. It’s a decent starting point for further adjustments. Measurements at rtings.com show that the X4 has a stronger response above 10 kHz; maybe a treble boost won’t be needed there. I didn’t try to flatten the bump from the ear canal resonance. As the foam tip starts to get worn out with more use, I expect that bump to flatten itself. Ditto with the relatively hot response at 10 kHz. I'm having second thoughts about that treble boost though. Measurements above 10k are really unreliable and unpredictable from coupler to coupler, so an adjustment there has no guarantee of translating well to an actual human ear. It really should be a shelf filter, which is a wide-band adjustment.

Starting from this preset, I dialed the response to suit my taste. I think the result sounds very, very good. The X3 isn’t dead silent like the my Airpods, but its self noise is low enough not to bother me when I’m listening to most music. It might become a problem with music with lots of quiet passages.

I also have a pair of the Jaybird Freedom F5, which originally cost $200 when they first came out. It actually feels more comfortable and secure to me than the X3. Its sound can be tuned similarly.

Apps for personalizing the sound can be really powerful tools, if they allow the user to make fine adjustments. After many cycles of listening and measuring, on and off over months, I now prefer the resulting frequency balance from the X3 or the F5 over any unequalized IEM that I’ve heard. Personalization won’t be the silver bullet for everyone, but for the rest of us, it can be a game changer.

I made a second attempt at a Harman-esqe EQ for the X3 with Comply Sport Pro tips:


This custom preset has more bass. Instead of using a filter band to boost a region in the top octave, I used it to flatten the bump from the ear canal resonance. Maybe it’s a little dependent on insertion depth, but it's not as sensitive to it as in the case with the silicone tips, where the resonance showed up as a tall spike. I don’t think this preset sounds as lively as the first one, but it might be a better starting point for most other people.

I made these two presets available in the MySound app. Search for "Yuri V". Just keep in mind that they're for the Comply Sport Pro, which are acoustically absorbent even even compared to other foam tips, and so damp treble resonances more. Maybe after trying again to get it closer to target, I’ll make another update here. If you have an X3 with Comply Sport Pro tips, maybe you can try out my presets and give me some feedback.

“The correlation between price and sound quality is close to zero and, slightly negative: r = -.16 (i.e. spending more money gets you slightly worse sound on average).” Sean Olive was referring to the chart below when he wrote that in his blog:


Let’s be fair here. They rate an IEM’s sound quality by how closely its frequency response follows their target curve. We also don’t know how they chose the models in the chart. They certainly don’t include the many, many kinds of $5 junk IEMs that you can buy at fashion discount stores like TJ Maxx and Marshalls, but we get what he’s trying to say. The subject of this review, the elusive Sony MH755, is sometimes the perfect embodiment of the idea.


The bundled earphone
You can’t drive to a store and buy the MH755 by itself. It’s the earphone that comes inside the box with the Sony MW600, SBH20, SBH50, and SBH52 Bluetooth receivers. Several eBay sellers sell replacement MH755 IEMs by themselves for $6 to $8 each. I have a strange love affair with bottom-of-the-barrel cheap IEMs, but I didn’t pay any special attention to the MH755 until I saw the measured performance at Speakerphone’s blog, ClarityFidelity.

Last year, I was experimenting with mods for bassy and v-shaped IEMs by injecting resin into their rear volumes in order to reduce their bass. The process reminded me of the MH755’s smooth measured response. Its midrange and treble follow the diffuse field and Harman target curves closely. I was thinking to myself that all I had to do was reduce the compliance of its rear volume and I’d have the perfectly tuned IEM.

Not long afterwards, I ordered three MH755 from eBay. The first thing I noticed was the short cable, which is designed to be used with a Bluetooth receiver like the SBH20. I use an extension cable with the MH755 when I’m not using it with one—a Radsone EarStudio ES100 in my case. The MH755 has a J-cord design, which means that the cable going to the right earpiece is longer than the one going the left. It’s supposed to go behind the neck on the way to the right ear. I sometimes find the arrangement annoying; I know it’s a deal breaker for some.


What a cheap IEM can sound like
I was surprised when I listened to them straight out of the wrapper. I didn’t hear the huge bass suggested by Speakerphone’s curves. Surely enough, my own measurements showed that my samples don’t have as much bass as the one measured at ClarityFidelity:


As you can see, my batch of three shows good consistency. I think they already sound very good without mods or EQ. The response at 200 Hz is only ˜1.5 dB up, relative to 1 kHz. At 100 Hz, it’s only 5 dB up. This kind of tuning is uncommon for an IEM that’s not trying to keep the bass flat, like an Etymotic.

The sample that Speakerphone measured is bassier: +7.5 dB at 100 Hz. This boosted mid and upper bass makes the lower midrange sound thick and congested. In fact, the Harman target curve has a low point at around 200 Hz. Many IEMs sound clearer when equalized to decrease the response in this region.


The MH755’s bass is a little elevated compared to the 2017 Harman In-Ear target. I prefer less bass than this, less than the target even, which is ok because the curve is just an average. The researchers who created the target prefer less bass themselves. On the MH755, it’s in the lowest octave where I especially feel the exaggerated weight, so I usually shelve the entire bass down with an EQ when it’s available. I still keep some of the bass above flat, but now it’s more in line with my calibrated home theater and two-channel stereo speaker systems (which have subwoofers that extend down to 20 Hz and 25 Hz, respectively).

The MH755’s broad peak at 3 kHz is about 11-12 dB up from 1 kHz, close to Harman’s original 2013 target for headphones. Since then, the curve has evolved, and after some feedback from listening tests around the world, the latest versions of the around-ear and in-ear targets now have the peak ˜2 dB lower. This doesn’t sound like much, but it’s in the area where human hearing is most sensitive. A 2 dB adjustment at 3 kHz is noticeable.

When I tested my own hearing, I discovered that the peak is centered at 3.3 kHz and that the resonance is very strong. An HD600 has a tall peak at 3.5 kHz that bothers some listeners, but it’s a good match for my hearing. Maybe that’s why the MH755’s 3k diffuse-field peak doesn’t bother me that much. It sounds close to what I’m hearing from my speakers. But I can imagine that, for some, the frequency response might give vocals an unwanted glare.

The treble has a smooth response and is free of nasty resonances. I’m hearing a small (seemingly) low Q bump between 7 kHz and 8 kHz. It’s the ear canal's half-wavelength resonance, so its frequency depends on the distance to the eardrum. Speakerphone seems to have inserted the MH755 deeper into the coupler, so the bump is at 9 kHz instead in his measurement. That seems to help fill in the dip in the response above 10k. I can’t insert the MH755 that deep in my ear canal even with the small sleeves.

To my ears, the overall balance is superb, and much better than many expensive headphones and IEMs. The MH755 doesn’t sound out of place in a comparison with well-known models that have stood the test of time.


Two of my three pairs have well-matched left and right channels, and the third is only off by 1-2 dB in parts of the treble. I hear a strong center image, especially with the first two pairs. Within its limitations, MH755’s stereo separation is satisfying and doesn’t call undue attention to itself. I’ve heard headphones and IEMs that play tricks with the phase to try to project a distinctive stereo image. But it’s like looking at shadows at the back of a cave instead of the real thing. It’s laughable when you compare them to a live performance, or even a good stereo or surround speaker system. I suppose it could sound pleasing if the trickery matches your personal HRTF. But it usually doesn’t work for everyone and the result often sounds like a cheap effect that negatively affects the tonality. The MH755 has none of that. It’s a simple minimum phase system.



Comparisons
I compared the MH755 to several other cheap IEMs. The local schools here have lost and found boxes, and they have lots of earphones because students lose them all the time. I got to measure several of them. Lost and found gives me an idea of what students are buying. There’s a lot of cheap Bluetooth buds, but I found two KZ IEMs among other Chi-fi models, so my guess is that they’re reading the buzz online. The rest are from my collection.

TL;DR version: I like the sound of the MH755 better than the stock sound of any other cheap IEM that I’ve tried.




Tuning the Response
This section is about EQ and basic mods. It’s mostly about controlling the bass response. January 4 Update: added information on the effect of output tube dampers and foam tips. Enter if you’re interested.


Fakes are out there!
Last year I took a risk and ordered three MH755 from an unknown eBay seller. I was lucky to get a good batch. A few months later, I decided to order two MH750, which have microphones and longer cables. I was hoping to get the same performance as the MH755. The seller had the words "100% Brand New OEM and Grade A Quality, NOT Generic, NOT Fake, NOT Replacement" in the page’s item description. Look what I got:


As you can imagine, they sounded horrible. I showed the seller this graph and one from the MH755 and explained to him that I was returning the products because they're obviously malfunctioning. It’s possible that these sellers don’t know any better if they’re selling a genuine item or a fake. Who knows where they’re getting their wares? They could have several suppliers themselves.

I don’t remember where, but a few years ago, I read that Sony changed the guts of the MH750 while keeping the exterior the same. The only visible difference is the white filter at the end of the nozzle on the older models instead of a black one. Perhaps I got two original MH750 instead of fakes. I should have taken a photo of the ones I returned.

Update: January 12, 2019


MH755 vs MH750
I ordered two MH750 from an Amazon seller after returning the bad ones from eBay. This time, I’m convinced that I have the real deal, but I’m unhappy with their boosted bass:


After rereading this post by Sead Smailagic, the designer of these MH-series IEMs, it seems that the typical MH750 has a bass response that’s 7.5 dB up, relative to 1 kHz.


This response is similar to the one on the MH755 measured by Speakerphone. On my MH750, the bass boost is closer to +9 dB. On one of them, the left earpiece makes a crinkling sound during insertion—a sign that the driver is bottoming out. My MH750 must not be properly vented in the front. Perhaps the bass can be decreased by restoring the front vent, if I can find it.

I prefer the sound that I’m getting from my three MH755 and their much more controlled bass. What’s not certain is if my samples are atypical and that most units measure like the one at ClarityFidelity, or worse yet, my MH750. That would be disappointing.


Low-cost and low-risk alternatives to the MH755
There are many inexpensive IEMs that follow a different target curve, and they have their many fans. But I don’t know anything dirt cheap that sounds like an MH755 out of the box, especially a good one with less bass. My problem with the MH755 is the risk of getting a fake when you order online. Even if you get a genuine unit, you still might get a sample with an obscene amount of bass. What are the alternatives?



Outro
Sony’s MH series of IEMs has been around for a while. They came out before the first Harman target was published in 2013. But somehow, with the MH755, Sony came up with a smooth frequency response that closely follows Harman’s latest crowd-pleasing target. (That is, if you consider a sample with less bass.) The performance doesn’t end there. It plays loud with low distortion, and it doesn’t suffer from a compressed dynamic range when playing back at high volume.

The Sony MH755 shows what’s possible from a dirt-cheap IEM. It puts to shame many, many products that cost several times its price. Why don’t manufacturers, including Sony, make IEMs like this easily available? I mean, look at the kinds of things that are being hyped as the best low-cost IEM:


It sounds worse than the cheapo IEMs from Panasonic, Philips, and Skullcandy. Maybe we should ask Massdrop and Sony to reissue the MH750/MH755, but with a few changes: less sample variation, longer Y-style cable with mic, and more colors available—all of this while keeping the price low. Call it the MH75x or something like that. If a cheap IEM with the performance of an MH755 ever becomes popular, it will become a benchmark, a well-known reference used to compare to other products. It will be easier to call out the many pretenders out there and force manufacturers to step up.

The 2018 iPad Pro lost the headphone jack but gained a USB-C port. Apple thinks that these new iPads ought to be used with Bluetooth headphones, so they sent Airpods to the bigger tech sites and YouTube channels along with the review units. Like it or not, wireless headphones are the future. But are still times when you want to use wired headphones. For example, if you want to use the iPad as an electronic musical instrument, you’ll want to use a wired connection for its low latency. Bluetooth headphones are tens, if not hundreds of milliseconds slower, and it makes the instrument feel a lot less responsive, even unplayable. I’m no gamer, but I imagine that some games could need the same kind of snappy response. For these kinds of applications, Apple now sells a USB-C to headphone jack adapter.

I’ll leave it to others to describe how well this dongle works with the new iPads. Macs don’t need it yet because they still have the headphone jack. A quick online search shows that some folks got this adapter to work with Android phones like the Google Pixel 2. I wanted to see how well it would work with Windows 10 PCs because so many of them have terrible built-in audio.

So far, I have tested the Apple USB-C adapter on a 2016 Lenovo Thinkpad Carbon X1 (running Windows 10 April 2018 update) and a Microsoft Surface Go (October 2018 update). I’m happy to report that the adapter just works. Windows won’t show it in its list of devices until you plug something in the 3.5mm jack. Here are the sample rates and bit depths that are supported:


I made some basic audio measurements just to make sure my units weren’t malfunctioning. TL;DR version: they’re probably working just fine. I’d like to see others make more careful measurements that go beyond just verifying that it works, like I did.



It sounds like?
As expected, the audio performance of this adapter in Windows is similar to that of iOS devices. I tried it with all sorts of headphones and IEMs, and I compared the USB-C dongle with the Lightning headphone jack adapter and also the last iPhones and iPads that still had the headphone jack. I’m pretty much getting the iPhone and iPad sound out of a Windows PC.

There isn’t any obvious distortion or noise. The only exception is on a Logitech UE600vi, where I can hear a very faint background hiss during quiet tracks. When I’m playing the digital zero wav file, the noise doesn’t change its character when I make the SSD busy or move open windows around. When it was released, the iPhone 6S had a problem with strange electronic noises that you could hear on sensitive IEMs when going to the task switcher screen and switching apps. That doesn’t happen with this USB-C adapter. (BTW, it’s not so bad anymore on the 6S since iOS 11. The noise still changes, but it’s quieter and less offensive.)


The Apple USB-C headphone jack adapter gets the same 1 Vrms maximum output as other iOS devices. (Assuming that your country isn’t limiting the volume.) It should be well established by now how loud it can get. On pop tunes like the one in the video pictured above, a satisfying volume level for me with the HD600 is 58 (out of a maximum of 100 in Windows). The HD600 is a walk in the park for this adapter because it draws so little current even at maximum volume. When level matched, it sounds about the same as a competent dedicated headphone amp. That is, if you can get it level matched. Some kinds of music are so quiet that you might not be able to get the HD600 loud enough on the dongle.

The UE600vi, on the other extreme, has a 9-ohm DC resistance. If you crank up the volume, the distortion will be a lot higher than on the HD600. Fortunately, you won’t be doing this on low-impedance IEMs because they’re usually very sensitive. The UE600vi, for example, has higher than average sensitivity, even among balanced armature IEMs. When I was watching movie trailers on YouTube, I had the volume level anywhere between 8 to 12 in Windows.


Comparison with the Google Headphone Adapter


Last year, Google removed the headphone jack from their Pixel 2 phones. Their accessory USB-C audio adapter was $20 at launch, but they lowered the price to $9 to match Apple’s Lightning adapter after strong criticism from Pixel fans. I can see why Google’s adapter was more expensive at the start. It has extra abilities, like 1.88 Vrms output—5.5 dB higher than either Apple adapter. And it manages this without any special cables. The Google adapter can also change its operating mode to 400 mV maximum when it detects a lower-impedance load like an IEM.

I ordered three Google adapters last year and tested one of them. I wrote about it in this thread: link. They’re quite competent, so it’s a shame that Google discontinued them. The new model is more expensive at $12, and reports online show that its performance is much worse than the original’s.

Apple’s and Google’s USB-C audio adapters sound similar. I don’t detect any obvious distortion or noise from either one, except for the noise on the UE600vi that I mentioned before. The character and the level of their noise is very similar, even down to the measurements. Both adapters’ output impedances are similarly low as well, so they deliver a flat response to almost all headphones. On my headphones and IEMs, they sound about the same.

The adapter also sounds like the headphone jacks on my Macs. Windows PCs, on the other hand, usually have worse onboard audio. My Thinkpad, for example, has noisy output. The Surface Go is quiet, but part of that is because of its high output impedance; I measured 46 ohms. An easy load like an HD6xx sounds ok, but it’s a disaster for some balanced-armature IEMs with multiple drivers.

Last year, I got the Google adapter to work on the Nexus 7 tablet and the 2017 Amazon Fire HD 10 using an OTG cable. Those two also have weak headphone amps with high output impedance. I have yet to test them with the Apple adapter. I’ll update this report when I do.

UPDATE: November 16


The mic input is good for hobbyists who want to make measurements
Windows 10 recognizes the Apple USB-C audio adapter’s microphone input, so any microphone that works with an iPhone should also work with Windows. The Dayton Audio iMM-6 measurement mic, for example, works just fine. A few hobbyists have used the iMM-6 or Chinese IEC711-compliant clones to measure the frequency response of their IEMs. Apple’s USB-C adapter works well with these. It doesn’t have the bass roll-off of the Google adapter. It’s also much cheaper and more portable than the Startech ICUSBAUDIO2D audio interface, while having a lower output impedance.


Even if the Google adapter rolls off the bass, it's something electrical that can be handled by a calibration curve, unlike the coupler's acoustic transfer impedance, where one calibration curve cannot fit all. The Apple adapter makes it one less thing to worry about.


What we have so far
$9 gets your Windows PC sound that’s good enough for all but the most difficult headphones and the most persnickety audiophiles. It’s also very light and portable. So what’s not to like? Maybe its build quality. Apple’s Lightning audio adapter has a bad reputation for being too fragile and easily damaged. It remains to be seen if this USB-C adapter will be the same. It has a similar construction.

It's a good thing, then, that it's cheap and easy to find in stores. Apple stores obviously have it in stock, but I suspect that Best Buy, Walmart, and Target will have it on their shelves soon. It’ll be strangely ironic if Apple’s USB-C dongle becomes the most common headphone adapter for Android phones. Maybe then, Android users will join their iPhone-using brethren in leaving one-star reviews online if this thing really is fragile. Courage!

Unless otherwise noted, I’m writing about my experience with firmware version 1.3.2 and iOS control app version 1.7.2 - 1.7.5. I first wrote about my experience with the ES100 and firmware version 1.2.1 in this thread, with some basic audio measurements. I’ve been using the ES100 on and off for the last few months, and I’ve been keeping notes. I put together my observations here.


I picked up the Radsone EarStudio ES100 back in May because it gives my portable devices a system-wide equalizer. If iOS and Android had something like Equalizer APO or eqMac2 without having to root or jailbreak, I probably wouldn’t have bought one. It’s the closest thing to an ultra-portable equalizer today. Remember these?

Well, the ES100 is the modern version, except that audio comes in via Bluetooth or USB instead of a line input. But it also does a lot more while being a lot smaller. Besides being a compact graphic EQ, it’s a versatile Bluetooth receiver that you can clip on a belt or even a headphone’s headband. If there are other super compact equalizers out there, Bluetooth or otherwise, I’d like to know about them.


What it is
The EarStudio ES100 is a very small Bluetooth receiver and USB DAC with headphone output. It has an all-plastic exterior that feels cheaply built at first because it’s very light. On the sides are volume up/down, next/previous track controls as well as a play/pause/on/off button. The rest of the controls are in the iOS or Android companion app.

It comes with the standard 3.5mm headphone jack, but there’s also a 2.5mm TRRS output that doubles the output voltage by bridging the amplifiers. On the front is a single ring-shaped status LED, and on one of the short sides is a micro USB port for data and charging.


Bluetooth receiver operation
The ES100’s Bluetooth connection has remained stable in the few months that I’ve had it. It has AptX and AAC, which are good enough for me. The likelihood of hearing offensive lossy compression artifacts is very low, especially with AAC. Other advanced codecs like AptX HD and LDAC are also supported, but low latency AptX LL is missing from the list on the product page.


When playing software synthesizers on an iPad, there is considerable lag. I don’t know how well the ES100 will work with some games. For video, the worst Bluetooth audio sync that I’ve heard is in the iOS YouTube app. There are a few audio/video sync test videos on YouTube that give you an idea of how much latency there is. With the ES100, it seems to be a little over 100 ms, which is not as bad as it is on my other Bluetooth devices: it’s in the same ballpark as the Airpods, and much better than my Bose QC35 and Jaybird Freedom. A work-around for iOS is to play the same YouTube video in Safari, where the audio is more in sync. Other iOS apps (like Netflix) fare better than YouTube. In Windows 10 or MacOS, YouTube audio/video is more in sync than in the iOS app.



The user has some amount of control over the latency, which can be reduced with a shorter buffer. The tradeoff is in reliability—with a shorter buffer, audio drop-outs are more likely.

The ES100 supports multi-point connections. It can remember several paired devices and connect to two of them simultaneously. It’s convenient because I don’t have to fiddle with the Bluetooth settings on my phone, tablet, or laptop when I want to switch between any two of them.

Another useful feature is the built-in microphone, which allows hands-free phone calls even with headphones that don’t have a built-in mic. The call quality depends on the placement of the ES100. It worked fine when it was clipped to the top of my shirt, but people I talked to said that they could hear me more clearly when I used the Airpods or the Jabra Elite 65t. When the ES100 was clipped to my belt, they said that they couldn’t hear me well at all.



The microphone can also be used to let you hear more of the outside. Radsone calls this feature Ambient Sound, and it comes in handy when you want to be more aware of your surroundings. It’s also convenient when you want to hear people and you don’t want to go through the hassle of removing and reinserting a highly-isolating, deep-insertion IEM like an Etymotic. The track advance button can even be configured to turn Ambient Sound on and off, so you don’t have to whip out your phone and look for the right page in the control app.


USB Audio
The ES100 works as a USB Audio Class 1.0 interface in MacOS and recent Windows 10 updates, so no drivers are needed. The operation is locked at 48 kHz sample rate and 16 bits per sample. In Windows 10, for example, you can’t select any other sample rate or bit depth. All other options are unavailable. This means 16/44.1 CD-quality audio (and any other format) is sample-rate converted to 16/48. The SRC math seems to be working just fine. I didn’t hear any problems playing back 16/44 audio, and my measurements back in May didn’t reveal anything out of the ordinary with 16/44 wav files.


Analog volume control and background noise level


As you can see from the screenshot above, there are two volume controls. Most of the time, it’s best to keep the digital volume at maximum, and just control everything using the analog volume control. The advantage of doing this: decreasing the analog volume also decreases the background noise coming from the audio jack. This keeps the background hiss low, even inaudible, when using sensitive IEMs. I tested a Logitech UE600vi, which has above average sensitivity even among balanced armature IEMs. With the analog volume at 0 dB, the background hiss is quite loud. But you wouldn’t be listening to music at a crazy level like that with the UE600vi. With pop music, a comfortable level for me is around -36 dB on the analog volume control. It’s higher for classical music—maybe as much as -24 dB for quiet tracks. At that level, I can’t hear any background hiss from the ES100, even with a digital silence wav file.

This makes the ES100 only the second Bluetooth device I’ve heard that’s completely silent. (I can’t hear any hiss from my Airpods either.) The physical volume up/down buttons on the ES100 control the analog volume, so it behaves differently in Apple devices than other Bluetooth headphones. In iOS, a Bluetooth device’s physical volume controls normally control the iPhone’s or iPad’s system volume. Not so with the ES100, and that’s a good thing because it gives you fine control: +/- 0.5 dB steps, unlike the very course adjustment in iOS.


Output impedance, distortion under load
When I measured my unit back in May, I found the audio output to be clean enough, even with the dummy load—16 ohms resistive, like the typical dynamic-driver IEM. I also verified the claimed 1-ohm output impedance. Since then, the app seems to have made another feature available: high current mode, which parallels two output stages and cuts the output impedance to 0.5 ohms.



With a high impedance load, the ES100 starts to clip when the analog volume is around +2 dB. With high current mode and a 16-ohm resistor on each channel, it clips earlier:




Here’s how much the output voltage drops when a 16-ohm load is connected to each channel:



Do the math: www.wolframalpha.com/input/?i=0.198+*+16%2F(R+%2B+16)+%3D+0.192

What’s strange is that I’m getting the same or better results when I leave it on normal mode. It looks like high current mode is on regardless of the setting in the app.


The main feature: the equalizer
With my collection of headphones and IEMs, the ES100 plays with low enough noise and distortion that I don’t have to worry about them. The only concern left is a big one, and it is really where the ES100 distinguishes itself from similar products. Many folks don’t want to admit it—that once you take care of noise, distortion, and lossy compression artifacts, it is the frequency response that determines the sound quality. The behavior in the time domain (e.g., the impulse response, the step response, square waves) is characterized by the system’s magnitude and phase response.

This is why I was looking for an ultra-compact equalizer that can give my mobile devices a system-wide equalizer without rooting or jailbreaking. Here’s a sample of what the ES100’s 10-band EQ can do:


This EQ can make many cheap headphones and IEMs sound good. Here’s an example: the Skullcandy Jib plays with low distortion, but its frequency response makes it sound like mud. Have a look: +8 dB at 200 Hz, and +11 dB at 100 Hz, relative to 1 kHz with the response still rising. It’s also missing a lot of information above 10 kHz:



Here’s the result with the ES100’s equalizer:



The Jib has low enough distortion that it can handle a boost in the treble, and with the sliders shown in the picture, the ES100 gets the Jib’s response close to the 2017 Harman In-Ear target. It’s a good starting point for further adjustments. Me, I prefer less bass and a little bit more above 10 kHz. But even before that, the result already sounds very good IMO—better than many IEMs that cost more than the ES100 itself. This shouldn’t come as a surprise to anyone; for years, people in this forum have been writing about how a carefully-tuned EQ can make dirt cheap IEMs sound like a million bucks.





Here’s another example—this time with a Sony MH755, a $6 IEM that already sounds very good without EQ:


I have several pairs of MH755 and MH750 IEMs and I chose one that has a little channel imbalance. In the graph of the equalized response above, notice that the left and right channels are more closely matched. The ES100 has a nifty feature that lets you trim each channel, and I used it effectively here:



The 10-band EQ works best with headphones that already have a smooth response, like the two IEMs above, or when there is a broad peak that lands close to one of the equalizer’s ten center frequencies. It's easy, for example, to give an Etymotic a little more bass. The EQ is more limited when you want to make the adjustment at a frequency that’s far from one of the graphic equalizer’s ten octave-spaced bands. For example, the Bose SoundTrue Ultra has a half-wavelength resonance that’s always at the same frequency because it has the same, repeatable, shallow insertion depth (if you could call it that). On my ears, it’s near 5.5 kHz—far from the ES100 EQ’s 4k or 8k bands.

Even if you could guarantee that the adjustment frequency is correct, the equalizer still can’t handle high-Q resonances, which look like tall, narrow spikes in the frequency response. Here are some examples that can't really be fixed by the ES100's EQ without the help of acoustic methods:


These two have the KZ house sound. The older KZ ATE also has that narrow spike near 8 kHz, depending on the insertion depth. For me this peak is not an artifact of measurement because I can clearly hear it in sine sweeps. Music sounds better when I use a parametric EQ to tame the bass and surgically flatten the peak with a narrow biquad filter. The ES100’s graphic equalizer, however, cannot effectively reduce the activation energy for these resonances.

I wish Radsone had included a parametric equalizer with the ES100. Fewer filters bands would be needed and better results can be had for most headphones. If DSP resources are limited, then maybe one could choose to turn off the graphic equalizer to use the parametric EQ.

It’s really a question of priorities. The ES100 has many features that made very little audible difference when I used them—the jitter reducer, the choice of reconstruction filter, the DCT filter, the DAC oversampling rate, and the additional codecs beyond AptX and AAC (although it is nice to have them). The EQ, on the other hand, has a huge impact on the sound. If there is an ES100 Mk.II in the works, my vote is for a more capable EQ, if it is impossible to update the current model’s firmware with one.


The crossfeed is different?
The ES100’s crossfeed has a strange implementation that’s not very effective. It seems to be very different from a Meier or Linkwitz crossfeed. When listening to speakers, each ear can hear a lot of low frequency information from the other channel. Therefore, the crossfeed should be stronger for low frequencies. This is not what I’m hearing from the ES100. Back in May, I got this response from the crossfeed:


Radsone’s implementation uses some inter-aural time difference, or ITD; i.e., the ES100 delays the sound on one channel before feeding it to the other one. I measured 2.5 ms before the crossfeed signal starts. You can figure out how far sound travels in that much time and decide for yourself if they got it right.

So many things that make little audible difference are tweakable on the ES100. So why not something that can easily be heard, like a way to select the crossfeed implementation?


Give me convenience or give me death
Once the new toy syndrome wore off, I took the ES100 out of the house less often, and usually found myself grabbing something more immediate: Airpods, Jabra Elite 65t, Jaybird Freedom, Bose QC35. Their sound isn’t as good, but they’re good enough for enjoying music. Even when the ES100 was already in my pocket, I still went for the Airpods or the Jabras. Just open the case, put them in your ears, and you’re good to go: no cables to unwind, no connectors to plug in, and no power button to turn on. And when you’re done, just put them back in the case.

Here’s one way to make the ES100 more convenient:



I’ve often used the ES100 like this at home. Just turn it on and put on the headphones. Easy. I was lucky that I had a short cable at hand to make it work. The clip doesn’t go around the SHP9500’s headband, but It clamps tightly enough to stay put as long as you don’t shake your head too vigorously.

It’s harder to find a short cable for other headphones:




A short cable and headphone mounting kit bundled in the box would have made the product more convenient to use. For IEMs, maybe a neckband or behind-the-head kit could work as an optional accessory.


Conclusion
The Radsone EarStudio ES100 is a useful little device. It’s also inexpensive. Audio performance is good enough for almost all listeners: It has low output impedance, it plays back audio with low distortion into realistic loads, its 2.5 mm jack provides double the voltage for insensitive headphones, its analog volume control makes the background hiss inaudible even when using sensitive IEMs, and it supports advanced audio codecs. But the killer feature, the one that improves the sound the most, is its equalizer.

It’s just too bad that they didn’t make it even better by offering a fully parametric EQ as well. The crossfeed could use some help too. Also, a short cable kit to make headphone use easy and convenient would have been a nice addition. Despite these complaints, I know of no other product that does the same things while being so small and affordable. I want to see what they have planned for the next version.

Every once in a while, the Jabra Elite 65t’s price drops down to $130, maybe lower. It happened during the two-week return period that I had them, so I asked Best Buy to match Amazon. The product page doesn't call these sport earphones, but the specs say they have an IP-55 rating, so I gave them a try. I guess they want to reserve the sport designation for the more expensive Elite Active 65t. I returned the Jaybird Run after 14 days and decided to keep these Elite 65t instead. Price was a factor in the decision.



How they work out during workouts

So far, they have stayed in place during my runs. It helps to have music playing because the earpieces don’t do a good job of decoupling the noise from your steps. My natural running style is mid-foot contact. Maybe if I were a heel striker it would be different. Even when walking, the sound of footsteps is magnified. The Jaybird Run does a better job here, especially with the Comply Sport Pro tips.

Maybe foam tips for the Elite 65t will help. Comply doesn’t make them specifically for this model. The generic Sport Pro with Smart Core tips fit, but they make the earpieces stick out of my ears too much. Comply Truly Wireless with Smart Core fits better; unlike the Sport Pro model, they don’t prevent the earpieces from going back into the charging case. I didn’t continue to use them because I didn’t like the resulting sound and couldn’t fix it with the app’s equalizer. Maybe in the future, Comply will make a foam tip specifically for the Elite 65t.


Update, January 12, 2019:


At the gym, the Elite 65t stayed in place during box jumps, burpees, barbell snatches, cleans and jerks, and handstand pushups. They didn’t feel as secure as the Jaybird Run, but they didn’t come flying out of my ears either. When I shake my head no vigorously, they stay put. Maybe problem is psychological, and with time, I’ll get used to them and just focus on the workouts.

BTW, as far as secure fit goes, the Jaybird Freedom F5 is the one that gives me the most confidence and has done so from day 1. I wear them with the cord over the ears and tightened securely behind the head. Secure fitting is a matter that is different from person to person, so of course, YMMV.

The exercises often take me a good 10m away from my phone. So far, the Bluetooth 5.0 connection has been stable. The Jaybird Run has only Bluetooth 4.2, and early reviews slammed them for connection problems. Jaybird must have had a lot of time to fix the firmware because I didn’t have any connection problems with them either. Range tests at rtings.com says that the Jabra does better, but neither model had an issue at the gym.


Some Useful features

Call quality is good—better than Jaybird. The four microphones are there for wind noise reduction. Like on the Airpods, calls just work with no problems. If known, the caller's name is announced in a rather robotic voice. There’s a feature called Sidetone, that allows you to hear your own voice through the mics. Talking while wearing IEMs still feels unnatural to me, so the experience is better with the Airpods. The Sidetone feature on the Elite 65t helps a little bit.

The Sound+ app can tell the microphones to allow some outside sound to come in. They call this feature HearThrough, and you can adjust its level. Jaybird doesn’t have this.

I find multipoint connections very useful. The Elite 65t will pair with up to eight devices and connect to two simultaneously. When I take them out of the case and put them on, they tell me, “two devices connected.” I don’t have to do anything to switch between my phone and whatever other device I have with me, tablet or laptop. The Jaybird Freedom F5 and X3 can do this too. The Jaybird Run cannot. I wish the Airpods could do this. Some intervention is needed when switching even between Apple devices.


About the sound


The default sound is fine. This is with the EQ set to flat. They sound like the common, inexpensive dynamic-driver IEMs that you can get from Sony, Philips, Panasonic, LG, or JVC. Here’s how mine measures with flat EQ:


This is a bit different from what rtings.com measured (unlike my Jaybird Run measurements, which have good agreement with theirs). They’re getting a huge 10+ dB spike around 10 kHz, and a bigger bass roll-off. I dislike headphones and IEMs with that sort of response. To me it sounds harsh. If I’m being generous, I’d call it fake sparkle that’s not supposed to be there. I listen to calibrated speakers at my computer, my home theater, and my keyboards, so it’s hard for me to get used to any headphone that does that. I’m going to guess that so many expensive boutique IEMs have something similar, to cater to corrupted audiophile tastes.

I’m glad that my pair of Elite 65t doesn’t have that peak, and I’ve verified my measurements by listening to sine sweeps. What I’m getting instead is a small bump around 7 kHz. The exact frequency depends on the insertion depth and the length of the ear canals. I can hear it with test tones, so it’s definitely not an artifact of measurement. To me, it might be small enough that’s it’s not too harmful to music. This is also where a proper foam tip for the Elite 65t might help.

I don’t know if the difference between rtings.com's pair and mine is in the hardware, or if it’s a result of a firmware update changing the default equalization. Either way, I’m glad that mine doesn’t have the 10k spike. I hope a future update doesn’t mess up the response.

The Jabra Sound+ app controls the built-in 5-band graphic equalizer. Every connected device gets system-wide EQ because the settings are stored in the earphones themselves. The Elite 65t retains the EQ even after powering down and powering up. Jabra's equalizer isn’t nearly as flexible as Jaybird’s 5-band parametric EQ, so it’s a good thing that my pair's frequency response isn’t too crazy. I played around with the EQ to get something close to the 2017 Harman In-Ear target:


Maybe I could have boosted 4 kHz more to get even closer. Anyway, this EQ has too much bass for me. With solo piano music, for example, the lowest notes have excessive weight. And the timbre is wrong. When you pound a piano's lowest keys, it should rumble and growl a bit. With this EQ, it doesn’t have as much bite, maybe because the notes’ harmonics are not in the right proportion. But I can see where this preset might come in handy. In a busy environment like the gym, the extra bass helps to mask the external noise. Also, there are some kinds of music that can benefit from such a boost. Here’s the other extreme:


Even with the 60 Hz and the 250 Hz sliders at their lowest positions (which looks like -6 dB), the bass isn’t completely flat. This might be the closest you can get to a diffuse-field sort of response. When I’m in a quiet location, my preference is for something in between:


Maybe it could use a little bit more at the frequency extremes to make it sound livelier. I might have made some incremental change since this measurement, but it sounds quite good to me. Pulling 250 Hz down to the minimum sounds a bit extreme, but after listening to it on and off at various levels, I definitely feel that 250 Hz at the minimum level sounds more natural, and that it very much improves the clarity of the music. It also makes the sound closer to what I’m hearing from my speakers.

Please note that this equalizer preset is my preference, and that I wouldn’t necessarily recommend it to anyone else. The ideal setting, of course, is program dependent, i.e., one curve might not work for all music. This is why I appreciate the convenience of having an easy-to-adjust system-wide EQ.

Jaybird’s parametric equalizer is more flexible than Jabra Sound+: each peaking filter’s center frequency and Q are adjustable. I can dial their frequency balance closer to my ideal response. A Freedom F5 or X3 with Comply foam sounds damn good when the EQ is carefully personalized to my taste.

The Jaybird Run’s frequency response can be similarly finely tuned. But I’d rather listen to the Jabra Elite 65t after equalization. The self noise is lower on the Elite 65t than on the Run. Its soft background hiss is still audible, unlike on my Airpods, which are dead silent. The Elite 65t also has lower measured distortion than the Jaybird Run. But that’s not where the Jabra wins big.

The Elite 65t supports AAC playback. The Jaybird Run supports only SBC, which is mandatory. SBC lossy compression artifacts are more likely to be audible—more than MP3 at the same bit rate. AAC renders audio better than MP3 and AptX, let alone SBC. On the Elite 65t’s product page, AAC playback is not a listed feature, but I can confirm AAC support. Here’s a screenshot of Bluetooth Explorer on a Mac Pro while playing music:


The Jaybird Run shows only SBC as the active audio codec even when AAC support is enabled. I wrote about my problem with this in my Jaybird Run review. It’s disappointing because their Freedom and X3 IEMs have AAC.


More about the Jabra Sound+ equalizer

The equalizer in Sound+ doesn’t have any markings on the vertical axis, so it’s just as bad as Jaybird’s EQ. I wanted to know more about the range of adjustment, so I measured the frequency response with each individual slider at the maximum and minimum settings, then subtracted the resulting curves from the one with the EQ set to flat. Here’s the result:


It looks like +/- 6 dB for each band. But look at the one for 7.6 kHz! The label is wrong. It looks like the center frequency is near 16 kHz, which would continue the pattern of having the adjacent bands two octaves away from each other. Or it could it be a shelving filter. If I didn’t make an error in measurement, the range looks like it could be wider than +/- 6 dB. If you were hoping to reduce the level of the peak near 7 kHz, please note that the slider labeled 7.6 kHz will have less effect than shown in the app.


Latency and video sync

Rtings.com measured the Elite 65t's latency at 224 ms. It’s practically the same on the Jaybird Run and the Bose QC35 II (at 220 ms each). The Jaybird Freedom and the Airpods do better at 179 ms and 168 ms, respectively.

None of them are suitable for playing software instruments in real time. When I tried the free (but good) Audiokit Synth One app for iPad, the differences feel bigger than these numbers. With the iPad’s built-in speakers, the latency is low enough that the notes play immediately when you press the keys. With the Elite 65t, it feels like half a second. The Airpods react much more quickly, but they’re still unusable.

I got mixed results when playing video in iOS. Audio-video sync depends on the app and how well it delays video to match the audio. iTunes and Netflix are watchable. Other apps don't sync as well; YouTube has one of the worst results. It might be a deal-breaker if you spend a lot of time there. But this is something that can be addressed in an app update.

YouTube was not as bad in Edge or Chrome on a Windows 10 PC. Search “audio video sync test” in YouTube for videos that can help you estimate the delay. It was close to 300ms on an iPad Pro and the Elite 65t. Get this: the lag is much, much smaller when you play YouTube videos in Safari! It’s practically in sync. The only problem with this is that the browser’s UI doesn’t disappear when you tell YouTube to play the video full screen.


Conclusion

Compared to what else is out there, it’s not bad for a true wireless sport IEM at $170. As of this writing, I’m unaware of anything cheaper with the same features and performance. It’s a better deal, of course, if your timing is good and you catch it when it’s $130.

The prices of these things have been fluctuating at Amazon. I asked Best Buy to match the price when it was down to $135. Early reviews were pretty negative because of Bluetooth connectivity problems. I guess they’ve had some time to make fixes with firmware updates. There haven’t been any connection problems for me, so far.

The earpieces stay put when I’m running. That’s what they were designed for, after all. Also, there hasn’t been a problem with jerky movements like cleans, snatches, burpees, and box jumps. I haven’t tried them with handstand pushups, which is when the Airpods fail to stay in. The Jaybird Freedom F5 passes all of these tests. (I wear them with the cord over the ear and then wrapped snugly behind the head. They’re now also cheaper: $50-$56 as of this writing.)

Some annoying things (compared to the F5 or X3): No multipoint connections, so it can’t connect to two devices at the same time. Also, no volume controls.


About the sound quality…

Flat EQ measures like this:




With the Comply Sport Pro tips, the resonances between 6 kHz and 10 kHz are more damped, so it’s easier to EQ with the MySound app’s 5-band parametric equalizer. Even then, the app is so fiddly and it’s really tricky to dial-in the desired sound. A measurement rig helps. My complaint from day 1 with the X3 and the Freedom F5 has been the same: there is no indication in MySound as to where you’re placing the EQ’s peak, or how many dB is the adjustment. The frequency markings on the horizontal axis don’t really help that much.

Eventually, I was able come up with an EQ preset that makes the response less V shaped, which is my preference:


I much prefer this user-defined preset, which is a good starting point for further adjustments. It seems to be somewhere between DF and Harman. When I program the same response on the Freedom F5 and the X3, the result sounds damn good—to my ears at least. But then again, my tastes may be unusual. I wouldn’t necessarily recommend this EQ preset to anyone else:


In principle, I’m against making large boosts like the one in the treble. But after flipping back and forth between having it and not, I decided that I like the sound better on the F5 and X3 with the peak left in. It looks like 5-6 dB. Maybe I’ll change my mind later. In case you were wondering, the IEMs remember the EQ preset even after powering off, and they're automatically applied to every connected device, even if that device doesn't run the MySound app.

Here’s the problem with the Jaybird Run: While the resulting frequency balance is very close to my preference, the result doesn’t sound nearly as good as it does on the Freedom F5 and the X3. The boosted treble only seems to expose the problem even more. What happens is that cymbals and high hats sound broken up, like crinkling cellophane wrappers. On the F5 and X3, they sound just fine—with a gentle and delicate shimmer, when it needs it.

It’s not just high-frequency content that’s affected: Long sustained vocal notes sometimes have a slight warbling sound with a crackling edge and a surrounding echo-like haze. The F5 and X3 don’t sound like this at all.

On the Jaybird Run’s product page, the specs show that it plays with more distortion than the F5 or X3. But there’s a more serious problem: AAC is not listed as a supported codec—only SBC. Indeed, what I’ve been hearing reminds me of low-bitrate MP3 lossy compression artifacts from a long, long time ago.

Before anyone calls me out on it, I must emphasize that these differences are slight. I didn't notice when I was doing my workouts. It wasn't until I got home and started to pay close attention with more quiet surroundings that I noticed anything. Even then, the easiest person to fool is yourself. There could be confirmation bias at work here. Also, it's really difficult to conduct a proper double-blind test with two different devices with different codecs.

Also, if there is, indeed an audible difference, I don't want to jump to conclusions that the codec is really to blame. In any case, I decided to investigate further with the Bluetooth Explorer for MacOS utility (available if you download Additional tools for Xcode).

Bluetooth Explorer on MacOS
You can enable AAC for Mac:


The Jaybird Freedom F5 plays audio just fine with AAC:


The Jaybird Run uses SBC even if AAC is enabled. And it has the same audible artifacts on the Mac that it has on the iPhone:


It's 226 kbps SBC even when AAC is enabled. You'd think it ought to be fine, because well-encoded MP3 sounds ok at that bit rate. If the problem is really with the codec, then maybe SBC doesn't render as well. In any case, it's not using the maximum rate of 328 kbps.

The Freedom F5 manages a slightly better data rate with SBC when AAC is disabled:



The competition
I’m currently evaluating the Jabra Elite 65t, which seems to be the Jaybird Run's main competition. They came out a few months after the Run, but look at the 65t’s advantages: AAC, multipoint connections, volume controls, Bluetooth 5.0 (vs 4.2), and a more sophisticated microphone system for better call quality and the ability to allow some outside sound to come in. Plus, it has lower measured distortion (at rtings.com).



The Jabra app’s EQ isn’t as flexible as MySound, but maybe the 65t won’t need as much adjustment. The default frequency response is better. It’s also something that could be improved with a future firmware and app update.

Update: My Jabra Elite 65t review

Conclusion?
If you’re on head-fi, you’re picky about sound quality. So if you haven’t bought the Jaybird Run yet, maybe wait for v2.0? Or try it and make sure the store has a good return policy. Maybe it will fit your needs. As usual, YMMV. In any case, I hope Jaybird address all of the issues I wrote about above.