The phrasing "intrinsically flawed" may NOT be the best way to say to comment on IEM frequency responses for a number of reasons.
1) everyone's ideal sound signatures is subjective individual preferences.
2) even if the "ideal sound signature" was standardized and an agreed upon goal, it is not always easily apparent where that line will be drawn on an individual's measurement graph, that line will look different on each different measurement rig.
3) everyone gets different fit with IEMs so the measured FR on the graph is not going to be identical to what someone hears when testing via FR sweeps or listening to music.
From my experience playing with measurement gear from IEMs, it is a very eye-opening experience as I saw in person how certain things measured can be applicable to my listened experience and how easily the measurement can be unreliable. The type of seal achieved & the insertion depth of the IEM in the tube would vastly affect the FR graphs I would generate and it is possible to get quite a bit of variability with measurements. Not only is it quite a hassle to get nice reproducible results during different measurement sessions, but I would find that the measurements do not always correspond what I've heard. Numerous explanations due to fit and seal, but there were definitely some measurements that I captured that did not correspond at all to what I would hear during a sweep test.
The other measurement reading error that you would not know unless you have played with measurement gear before is that he is using a Vibro Veritas coupler (it is basically a black tube that you stick the IEM into and place gum around it to make a seal) which is really only accurate to 100Hz to 10kHz. While it may still capture relative peaks & valleys before and after that, the actually dB does not correlate well to the rest of the graph. Also, for the treble region, it is not uncommon to have artificial peaks/valleys that is an artifact from the measurement system which does not actually correlate with you do a listening FR sweep test. I actually have the same measuring set-up at home for IEM testing. So your comment talking about the vast change in dB between 20hz to a random valley in 16kHz does not actually correlate to anything in real life, the measuring coupler is not accurate in either of those FR regions. There is not that kind of perceived change between the overall bass and treble region. When even looking at measurements from more professional gear like by Tyll, you can see on the TOTL flagships over-ears such as the Utopia you can calculate out a 30+ dB sonic difference if taking the highest point in the bass region and comparing it to the lowest peak in the treble region. That is not the right way to read measurements.
Some helpful information regarding reading FR measurements:
https://www.innerfidelity.com/content/headphone-measurements-explained-frequency-response-part-one
Measurements are definitely helpful for showing the comparative degree of difference between different IEM models, but it is too extreme of a stance to call a headphone "intrinsically flawed" based purely on FR measurements comparison. Especially when the person who actually took the measurements & did the listening comparisons (
@csglinux) had the following the say about the comparison:
While the xelento's are definitely bassier than the ER4XR that does not mean that its sound signature is "intrinsically flawed." Other TOTL IEM with similar degree of change between its overall bass and midrange on measurements include the T8iE, campfire atlas, empire ears legend x, and noble K10. In fact, referencing all of csglinux's comparative measurements, you can see that some good flagship reference points that are praised for realistic overall tonal balance: the electrostatic shure KSE1500 and campfire andromeda is measured to be 8db at 100hz (with zero at 1k). So the xelento being at ~12db at 100hz is definitely not a crazy extreme bass boost as it is a positive 4dB boost from a good flagship reference point. The ER4XR is at ~4dB at 100hz, so that is negative 4dB from those reference points, does that mean that the ER4XR is crazy bass deficit? No, it is not an extreme change, but 3-5dB change is definitely audible. In fact the change between the AKG 701 vs the Massdrop produced K7xx is approximately a 3-5dB bass boost and you can hear the difference in a side-by-side comparison, but in no way does it completely mask out the other frequency responses and cause an "intrinsic deficit."
The Campfire Andromeda is often used as a benchmark for well-balanced sound signature and you can see in his actual comparison graph, there is a overall ~8dB boost in bass from 1k and the graph shows a treble dip to -17dB in the treble region. Just reading the graph, you would assume that they sound like bass monsters with severely hampered treble. But the Andromeda actually sounds like a overall bright pair of headphones. The measurements are an useful analytical tool, but they are not actually 100% reflective of what is heard as based on different compensation curves being applied and the different measurement set-ups. The degree of the peak and valley is not always reflective of what is heard.
Here is the xelento vs andromeda from csglinux's review:
https://cdn.head-fi.org/a/9953731.png
And here is an example of how different the FR of the andromeda can look with a different compensation and measurement rig:
https://i.imgur.com/2ZeXxyi.jpg
And here is the xelento on that different measurement set-up:
https://i.imgur.com/IgbmqE9.jpg
credit to
@crinacle for those comparative measurements.
When someone talks about the ideal FR frequency response, automatically the harman target frequency response developed by Sean Olive comes to mind. The experiment is to find what measured FR most closely correlates to what listeners perceive to be tonally flat. So while the ideal loudspeaker curve is either a flat straight line or downsloping straight line depending on who you ask, when you playback that same FR curve through over-ear (supra-aural) headphones vs on-ear headphones vs in-ear monitors, the actually frequency response that our brain perceives is not the same flat line anymore. In fact, the ideal IEM frequency response that they found in 2017 is a 8db increase at 20hz. compared to 1000hz. (source:
https://www.innerfidelity.com/content/harman-tweaks-its-headphone-target-response). Now doesn't the Andromeda and KSE1500 FR responses make more sense? A bass-boosted measured frequency response does not necessarily mean that it sounds unrealistic as the IEM format struggles with creating that realistic visceral subwoofer sensation that loudspeakers can create and often requires elevation of the lower frequencies to achieve that perceived effect.
Lastly, when someone talks about FR curves, you need to be familiar with the idea of equal-loudness compensation curves. The concept is that if you were to play a steady tone across the frequency response at a specific set decibel without compensation, we would actually hear it get louder when we approach the midrange. This is due to how our ear autonomy and the sensitivity of our hearing to different frequency responses is variable. So a pure tone played at a constant decibel would actually change in perceived loudness as we move across the frequency response. Humans are naturally more sensitive to the mid-range. The original equal-loudness curves was the Fletcher-Munson curve. It is a compensation curve that shows on average for our hearing to perceive something to be tonally flat there is actually a 50 dB drop moving from 20Hz to 1,000Hz.. But due to individual's having different sensitivities to different regions of the FR, there is individual variation what people can perceive as a flat frequency response without neither person being wrong. There are actually objective reasons for differences in subjective impressions in the world of audio, and I do think it is important to respect that fact.
I am uncertain exact what you are talking regarding only being able to hear a 37dB audible difference, possibly referring to the phenomena of auditory masking(?), but in that case the degree of different between the required loudness of the original signal and masking signal is variable depending on the sounds. A signal that will masking of 37dB of difference in loudness would be quite loud. 30-40dB is the sound of a fan, jumping to 60dB is talking, 110 is loud concert. Even playing music from a loudspeaker at normal listening volumes ~80dB, you can still hear a fan in the background. If there is no background noise (dependent on personal hearing abilities as well), the average person can be reasonable expected to pass a blind test with a 3dB volume difference (you can try the blind test here:
https://www.audiocheck.net/blindtests_level.php?lvl=0.5). There are people who claim that they can do better. So the phenomena of auditory masking depends on a variety of factors, but it basically states that if there a specific signal and another background noise at a specific frequency response and certain decibel is added, it has to be a certain degree of decibels relatively louder, the more background noise the louder it have to be with that signal to keep it from being hidden. Aka if your air conditioner is louder than your fan, it is harder to hear it fan, but to mask the fan completely with a different signal, I think the degree of jump required is going to be beyond the scope of variation of dB when listening to a pair of headphones at a safe volume. Even listening to a loudspeaker at average volume ~70 to 90dB, you can still hear quiet background noises. In recorded music, the variation of loudness between sounds is already prefixed, the masked signals already masked, and typically that variation of loudness within a track is much greater degree than the dB difference within the FR response of a headphone.
Of course, there can definitely be such an extremely skewed sound signature to the degree where you have to shake your head wondering what the designers were thinking, but this is not the case in the frequency response measurements of the xelento. Without understanding a lot of complexities behind measurements, we can often make mistaken assumptions regarding headphones. Despite my experience with headphone measurements, I still even make these types of mischaracterizations jumping to a conclusion too quickly based on skimming a FR graph that cause inaccurate expectations prior to doing listening tests. It is important to have some background knowledge regarding the measurement equipment used & limitations, type of compensation used, and what an ideal FR would even look like in that particular measurement system. The last point regarding the ideal FR is often personally extrapolated based on other headphones you've heard that has also been measured in the system so you can judge what degree of correlation the peaks and valleys relate to your personal hearing sensitiveness, but the idea that there is a fixed objective universal ideal neutral FR graph that fits everyone's ears is not realistic as there is individual variables that factor into what is a neutral-sounding headphone FR.
This turned out to be a much longer & more involved post than I was planning, but hopefully, it sheds some light on measurements & why I think that you are making a mistake calling the xelento intrinsically flawed. The xelento's bass response is nowhere near exaggerated to such the levels to cause significant auditory masking.
![s-l1600[1].jpg](https://cdn.head-fi.org/a/12097776.jpg)
![s-l1600[1].jpg](https://cdn.head-fi.org/a/12097763.jpg)
![A9d9f5c47dd5c4a08bcdb5abdf700906ae[1].jpg](https://cdn.head-fi.org/a/12097732.jpg)
![A9d9f5c47dd5c4a08bcdb5abdf700906ae[1].jpg](https://cdn.head-fi.org/a/12097731.jpg)
![AaBZZ8w[1].jpeg](https://cdn.head-fi.org/a/12097693.jpeg)
