staticV3
Master Contributor
- Joined
- Aug 29, 2019
- Messages
- 8,025
- Likes
- 12,873
Considering the driver setup, distortion is kept well under control I think.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
OP
- Thread Starter
- #62
The IEM Harman Target 2019 sounds "off" to me. Is it just me?
I recently purchased an Etymotic ER4XR. If you do so, you get a certificate with measurements that the drivers are matched within 1 DB. They already sounded pretty great to my ears out of the box. :) Still, I was curious how they sound if I EQ them to the Harman Target. I used the Harman tuning...
www.audiosciencereview.com
OP
- Thread Starter
- #63
Now from these measurements again the Nova looks more V shaped, more Harman like, like the original Zero with more sub bass.
If you insist on continually beating this dead, decayed, and by now forgotten horse, why don't you point people towards something useful, like @jaakkopasanen 's comparison of couplers instead of crying "inaccurate", "fake" and "knock-off" every time Harman and @crinacle come up coincidentally?
Jaakko: "The average error is almost zero across the board and the standard deviation is in check"
"Right hand graph shows what would be the compensation (or target) curve for the measurement system. This is done by adding the average error to Harman over-ear 2018 target."
Below 10kHz it's within the IEC tolerance itself (see page 6 of the GRAS instruction manual) as crinacle himself notes. In other words, if you had a real GRAS coupler, not a "fake" one, which seems to be your main sticking point, which had the same impedance as crinacle's, GRAS themselves would say it's fine.
If that's not good enough for you, then instead of crying "fake" you need to be honest and say instead "the IEC tolerance itself isn't tight enough for me personally to feel happy comparing measurements made on different rigs to the harman target".
Oh neigh neigh neigh, this horse is well and truly alive and kicking. I suggest looking more carefully at the actual data instead of blindly believing without question the words you read from sources like a YouTube influencer and in turn dodgy manufacturers of knock-off acoustic couplers. Do you realize how high the IEC tolerances actually are? This should educate:
So the IEC tolerances are up to 10 times the manufacturing tolerances of the GRAS couplers as used by actual professional acoustic engineers who measure headphones for a living like Oratory. Now for a specific example. Here's the Truthear Zero as measured by Crinacle with his knock-off '711' coupler, compensated to the Harman target to show the perceived tonality to the majority:
Looks fairly flat, close to 'full-Harman' as Crinacle marketed it as, right? Now take a look at Oratory's measurement using a genuine GRAS coupler (as Harman did when developing their target of course):
Now we see the reality, a somewhat v-shaped tonality, just as many have heard them as and wrongly attributed to qualities of the Harman target itself, all due to inaccurate measurements from knock-off couplers presented to and swallowed down by the masses as fact. By all means, if you want to play science-cosplay behind closed doors (whatever you're into, no judgement), ogle at and fiddle away with YouTube measurebaters' squiggles from their fake knock-offs, but as I said, in order to accurately judge Harman compliance, and therefore perceived tonality to the majority, these fake-science measurements need to be roundly ignored, and only genuine GRAS measurements considered.
Last edited:
MacClintock
Addicted to Fun and Learning
- Joined
- May 24, 2023
- Messages
- 538
- Likes
- 984
No, not for IEMs (as there was done a lot less research than for headphones). And if it were, just because it is a standard everybody has to follow it? If it sounds too bright to me, I have to endure it because they say so?
Last edited:
MacClintock
Addicted to Fun and Learning
- Joined
- May 24, 2023
- Messages
- 538
- Likes
- 984
Presented like this (and using the bass boost adapter), it is clearly shown that the Nova and the Red only deviate significantly above 1kHz, the Nova being closer to Harman, but like many, I prefer less energy in the ear gain region as the 2019IEM target, so for me the Nova offers no improvement, just up to now the best combination of price versus Harman compliance.
Edit, if one looks at the 5128 measurements from crinacle, they look almost completely identical, i.e now the Red without bass boost. And the difference in the treble is much less pronounced. A bit hard to make a judgement from the combined measurements.
Last edited:
Hey look, that test is as Harman-endorsed as it gets. The parameters selected by Dr Sean Olive himself, watermarked by Harman International, performed on a genuine GRAS system; if it's the topic is on Harman adherency then that's the highest authority anyone could appeal to.
Unless of course, Harman/Olive now finally isn't immune to criticism and we can talk about the shortfalls in their work and testing methodology...
There's no pleasing him honestly.
@GaryH I've put off replying because I believe the research I shared by Jaako settles it and I don't think your response changes anything, but this is mainly because I don't understand some aspects of it. So I'll bite:
Could you please explain what you mean by "the IEC tolerances are up to 10 times the manufacturing tolerances of the GRAS couplers"? I've looked at the table from oratory1990 you said "should educate", and I've looked back over Jaako's comparison, and I don't still don't see what you're referring to.
Secondly, could you please let me know the source of what you share as Oratory's measurement of the Truthear Zero? I wasn't aware he had measured it at all, I can't find it in his reddit index, and upon looking for it I couldn't find it anywhere. In https://kurin.squig.link/data/ all I can find is Resolve's measurement which, incidentally, looks very much like Crinacle's measurement on his imposter coupler.
Finally, I'd really like to understand the significance of the single measurement of an unknown IEM on an unknown coupler comparing 2 different brands of IEM tips which you share above, and which you combine with speculations but not measurable data ("potentially decreasing the upper mids", "likely usage with with a low output imedance")? This seems strange to me considering your general insistence on tight tolerances and genuine equipment, especially in the face of your complete rejection of Jaako's comparison above? Alternatively, why do you believe Sean Olive shared the measurement with foam eartips, as I can't see how it would benefit him to do so if it significantly differed from measurements with other tips?
That was posted on Twitter, interesting bit that GaryH cutoff was that Oratory himself said they broadly matched Harman with "1% of V" lol.
I hate that the thread is getting derailed again. But GaryHs mystery data doesn't match Oratory1990s twitter post. Verify your data everyone. Continue the discussion in a suitable thread, please: https://www.audiosciencereview.com/...get-2019-sounds-off-to-me-is-it-just-me.38113
Last edited:
It's not about pleasing me, it's about maintaining scientific standards and accuracy, on a science forum.
Happy to answer your questions.
The table shows the IEC tolerance is up to 2.2 dB. As Oratory says underneath this, the GRAS coupler manufacturing tolerances are 0.2 dB.
2.2 / 0.2 = 11 ≈ 10
(I was being conservative)
(Orange curve #4 with length mode peak at 8 kHz, not averaged, undamped GRAS RA0045 coupler, so matching Crinacle's measurement conditions, unlike others you'll see.)
I didn't 'cut that off'. What you have cut off however is what he actually said was they have "maybe 1% of V-shape in the mixture" (my emphasis), which is obviously not a serious quantitative calculation, but rather a casual turn of phrase with no claim to accuracy or degree of audibility.
That was for illustrative purposes (from a comment made last year, not long after the Zero was released), showing the potential difference foam tips can make. That's all that's needed to qualify it as a confounding variable. If someone is claiming a known potential confounding variable is not a confounder in a specific case, the onus is on them to prove this, not the other way around. Anyway, we now have data showing differences in FR with foam vs silicone tips specifically for the Zero, so that's moot. The 30 ohm output impedance used in Sean Olive's measurement is obviously a confounder, clearly measurable and its effect calculable from the Zero's impedance graph (which Amir has measured).
I don't reject it, just misinterpretations of it. The mean error (dark blue line) in the central plot of the image you posted is not what really matters, it's the standard deviation of the error (light blue shaded area) that's important, along with the leftmost plot showing all the individual errors. If you look carefully at the standard deviation of the error, you'll see this isn't even within IEC tolerances, and as Oratory says the latter are already pretty loose and in turn significantly larger than GRAS manufacturing tolerances. Then there's the obvious fact that the error for many individual IEMs in the leftmost plot significantly exceeds the standard deviation of the error, the Truthear Zero being another such example:
This all makes it abundantly clear how inaccurate Crinacle's knock-off coupler is. It's no wonder people are confused in this very thread about the Nova's actual response when they're looking at measurements from a coupler this inaccurate, along with a proliferation of just as inaccurate (but in different ways) measurements from other wannabe youtube influencers and their knock-off couplers. None of these measurements can be trusted, and should be rightfully ignored.
Likely because they're easier to get a seal with the artificial pinnae used for that measurement.
Last edited:
For the love of science man give it up. You need to learn the difference between a tweet and actual scientific research. And the difference between test conditions that aren't the same, and shortfalls in testing methodology. There are multiple test conditions that are different from yours with the Zero measurement by Dr Olive you posted which have known effects on frequency response: foam not silicone tips, a pinna is used, 30 ohm output impedance, etc. This is a science forum. An apples to oranges comparison will not be entertained on here by anyone with an understanding of basic scientific principles like controlling for confounding variables between tests. Nothing you say, especially not more desperate, flailing, failing attempts to further denigrate Dr Olive and his in-ear research, will change the data I've presented above patently showing the inaccuracy of your knock-off coupler. Just accept your '711' measurements are flawed due to this clear inaccuracy, take it like an adult and move on.
If the data seems so conclusive, why the need to use language that is both pandering and rude? Occams razor would suggest the opposite stance when it comes from a strong position.
How about this. If we're on the topic of accuracy (or lack thereof), then IEC60318-4/711 would be patently less accurate than the ITU-T Type 4.3 standard when it comes to emulating the acoustic impedance of the human ear. I know that there was a point on this forum where every 5128 publisher would be villified for even attempting to start a database on the new standard, but now even Harman sees merit in repeating the AE/OE 2015 research for in-ears on the 5128.
And even Dr Sean Olive admits as much, saying that the 5128 provides better human accuracy whereas the traditional 711 provides better precision
So since you're going on and on about accuracy, how about we look at 5128 measurements of the Nova instead?
And as long as that's exactly what you're doing in regards to Jaako's comparison, I'll re-quote that post.
"You must think you're clever throwing this graph around but it doesn't show what you hope it shows.
This is the exact same exercise, but this time comparing CSGlinux's database, which uses a GRAS RA0045 coupler, to Oratory's, with the following list of IEMs :
I've tried to eliminate some IEMs for which different configurations are available, and for which I had doubts which one was used, and probably didn't include them all, but I considered different tips and insertion depth fair game since that's also exactly what happened in the original comparison. There's probably a few mistakes in there as well, but so will be the case in Jaakko's comparison, it's to be expected with this approach.
I don't really know how Jaakko's comparison was normalised, so I've gone for two normalisations, one at 500Hz, another one across five octaves centred at 500Hz.
That would be the average difference, using REW :
I'd prefer to know more about how Jaakko performed the calculation for Crin's average difference before directly comparing them, but so far the magnitudes involved are basically similar.
Are the differences for CSGlinux's coupler representative of the actual difference between their RA0045 couplers ? Most likely (hopefully) not. There are just too many variables at play here, and large, bird's eye view database comparisons of this kind are quite likely inappropriate for that task beyond a certain degree.
Performing my own measurements has not only been quite helpful for my own benefit, it's also helped me a lot to learn how to interpret graphs, so maybe you should go on Aliexpress, purchase a clone coupler, and regardless of whether it's a truly good one or not, get some first hand experience by being confronted to all of these variables.
Otherwise next time, perhaps drop the pretend science posturing and make a modicum of effort to actually engage with, read and understand what you yourself quote. There's just too much easily obtainable data graciously provided by many parties out there to be that casual with it and with such an unconstructive attitude."
What you're doing by throwing around the comparison between Crin and Oratory's measurements of the Truthear Zero, in addition to not understanding that this comparison also has "confounding variables" that weren't controlled for (it's inevitable in these situations) - a lack of understanding that probably comes from a lack of experience about measuring IEMs, hence why I'd stick to recommending you to purchase a clone coupler and start discovering all of these confounding variables at play -, is the equivalent of doing this :
And claiming that CSGLinux's coupler is inaccurate as a result.
Except that these were both measured on a genuine Gras RA0045.
You'll need another method than Jaako's - and even more so simply throwing graphs of individual IEMs around - to have some idea about how accurate Crin's coupler is, beyond what these methods can realistically tell you.
You know all those people who constantly have to remind you how smart they are and invariably turn out to be imbeciles ?
On ASR that would apply to those who pontificate about "science-this science that".
OP
- Thread Starter
- #80
Nova is now released, instant buy for me of course. 
Last edited:
Similar threads
