• 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!

The IEM Harman Target 2019 sounds "off" to me. Is it just me?

isostasy

Senior Member
Joined
Oct 19, 2022
Messages
354
Likes
628
I had some responses in mind, but I'm checking out of this argument and leaving this here:
Back on topic, is the Harman target used in AutoEQ even correct? One outcome of the above exchange is that, in looking at the Truthear Zero on AutoEQ, I noticed it overshoots almost the entire bass shelf and only drops a ~2dB (from sight) below 20Hz. Every other measurement, including Amir's and crinacle's, show it following the target then starting to drop off around 50Hz. Looking closer, the bass shelf in the AutoEQ png is 5dB above 1kHz, whereas on crinacle's tool and Amir's target it is more like 7-8dB. I looked at some other IEMs on AutoEQ and it's the same, the target is clearly different.
 

isostasy

Senior Member
Joined
Oct 19, 2022
Messages
354
Likes
628
AutoEQ uses a modified version of the Harman target with 2dB less deep bass, for both headphones and IEMs.
Thank you! That makes perfect sense on its own, but I've further confused myself by going back over the AutoEQ documentation:

'oratory1990 measurements have been done on Gras 43AG and 43AC couplers, the same which were used to develop Harman target responses by Olive et al. and therefore use Harman target responses for the equalization targets'

...

'All of the results use frequency response targets that were specifically developed for this project except oratory1990 and Crinacle's IEM measurements which use standard Harman targets. The target curves were developed by calibrating measurements against reference measurements by oratory1990 and Crinacle (IEMs) and modifying the Harman 2018 over-ear and 2019 in-ear targets with the calibration data.


None of these targets have bass boost seen in Harman target responses and therefore a +4dB boost was applied for all over-ear headphones, +6dB for in-ear headphones and no boost for earbuds. Harman targets actually ask for about +6dB for over-ears and +9dB for in-ears but since some headphones cannot achieve this with positive gain limited to +6dB, a smaller boost was selected.'

I think I'm being stupid but I'm having a hard time following this: is it the last bit which implies a 2dB reduction, because he's saying there's a 4dB boost instead of 6dB?
 

JanesJr1

Addicted to Fun and Learning
Joined
Jun 9, 2021
Messages
505
Likes
450
Location
MA
I have ER4XR, ER3Xr and ER2XR. I happily use Oratory 1990 as a first-cut EQ for my other headphones, but not for the Ety's, where I agree they just don't sound right, Too bright, at least. My own EQ is waaay off the Oratory target.

The ETY's were my first audiophile HP/IEM's. The more experience I've had since then with other gear, the more I tend to doubt the Ety's as well as the target. They are resolving (except for the ER2), which I appreciate, but are bass-shy and lack the better tonality I've been able to find with my headphones and Moondrop IEM.
 

markanini

Major Contributor
Joined
Feb 15, 2019
Messages
1,719
Likes
1,769
Location
Scania
Common claim:

* Each individual has a unique ear canal and pinna, therefore everybody listening is unique and can't be shared.

In reality and because evolutionary reasons people do not listen significantly different. The notion that everybody lives in their own acoustic universe is nonsense. There's always the extreme outlier but that's rare.
This is a super important point. It's also important to avoid any conflation with personal fit.

I strongly suspect there are a lot of audiophiles who prefer a more analytical sound, which would deviate from the balanced but bassy harman target.
Are you sure that characterizes how audiophiles perceive Harman? Because a lot of popular IEMs on forums have more of a downward tilt compared to Harman.
 
OP
D

Dazerdoreal

Active Member
Joined
Oct 5, 2022
Messages
212
Likes
216
I have ER4XR, ER3Xr and ER2XR. I happily use Oratory 1990 as a first-cut EQ for my other headphones, but not for the Ety's, where I agree they just don't sound right, Too bright, at least. My own EQ is waaay off the Oratory target.

The ETY's were my first audiophile HP/IEM's. The more experience I've had since then with other gear, the more I tend to doubt the Ety's as well as the target. They are resolving (except for the ER2), which I appreciate, but are bass-shy and lack the better tonality I've been able to find with my headphones and Moondrop IEM.
In my opinion, you get the best results for the ER series when you use Oratorys measurements but not his EQ.
Oratory took into account that the resonance peak is not the same in everyones ear, so he ignored it and used a wide filter instead (which makes sense).

What I did and what I recommend to do is:
- On the AutoEQ page you can find Oratorys measurements as a csv file (https://github.com/jaakkopasanen/Au...atory1990/harman_in-ear_2019v2/Etymotic ER4XR). I loaded them into Squiglink (https://precog.squig.link/).
- Then I used a tone generator to find my own resonance peak. I was lucky in this case as my perceived resonance peak was almost exactly the same spot as in his measurements.
- So I used the AutoEQ function of Squiglink to EQ the ER4XR to tune it directy to Harman 2019 (and other targets I liked). If the peak was elsewhere I had to adjust it.
- I didnt use the Parametric EQ-File from the AutoEQ homepage. Just as Oratory, AutoEQ also cannot know where my personal resonance peak is.
 

isostasy

Senior Member
Joined
Oct 19, 2022
Messages
354
Likes
628
In my opinion, you get the best results for the ER series when you use Oratorys measurements but not his EQ.
Oratory took into account that the resonance peak is not the same in everyones ear, so he ignored it and used a wide filter instead (which makes sense).

What I did and what I recommend to do is:
- On the AutoEQ page you can find Oratorys measurements as a csv file (https://github.com/jaakkopasanen/AutoEq/tree/master/results/oratory1990/harman_in-ear_2019v2/Etymotic ER4XR). I loaded them into Squiglink (https://precog.squig.link/).
- Then I used a tone generator to find my own resonance peak. I was lucky in this case as my perceived resonance peak was almost exactly the same spot as in his measurements.
- So I used the AutoEQ function of Squiglink to EQ the ER4XR to tune it directy to Harman 2019 (and other targets I liked). If the peak was elsewhere I had to adjust it.
- I didnt use the Parametric EQ-File from the AutoEQ homepage. Just as Oratory, AutoEQ also cannot know where my personal resonance peak is.

Do you mean the resonance peak between 9 and 10kHz? I would suggest if you're getting a resonant peak here you're not inserting them deep enough. I've found it mostly unhelpful to go by any measurement of an Etymotic because they are designed for a deep fit which eliminates resonant peaks in the lower treble, whereas they will be fitted to a coupler with the intention to reproduce the peak at a particular point (e.g. 8kHz for crinacle's database) in order to improve comprehension.
 
OP
D

Dazerdoreal

Active Member
Joined
Oct 5, 2022
Messages
212
Likes
216
Thank you for your suggestion. I tried to insert them even deeper but I still heard a peak around 9k without EQ. If I tried even harder it would become painful and I had to worry that the tips get stuck in the ear canal.

But that is not a big deal, I am very satisfied with the sound as it is right now (I usually use the "Banbeu target"). :) And I am certain this is not the reason for my concerns with the IEM 2019 Harman Target.

I think that I already insert them very deeply. I sleep with foam ear plugs every night so I am used to the deep ear fit procedure (I have a sensitive sleep).
 

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,348
Likes
1,804
In my opinion, you get the best results for the ER series when you use Oratorys measurements but not his EQ.
Oratory took into account that the resonance peak is not the same in everyones ear, so he ignored it and used a wide filter instead (which makes sense).

What I did and what I recommend to do is:
- On the AutoEQ page you can find Oratorys measurements as a csv file (https://github.com/jaakkopasanen/AutoEq/tree/master/results/oratory1990/harman_in-ear_2019v2/Etymotic ER4XR). I loaded them into Squiglink (https://precog.squig.link/).
- Then I used a tone generator to find my own resonance peak. I was lucky in this case as my perceived resonance peak was almost exactly the same spot as in his measurements.
- So I used the AutoEQ function of Squiglink to EQ the ER4XR to tune it directy to Harman 2019 (and other targets I liked). If the peak was elsewhere I had to adjust it.
- I didnt use the Parametric EQ-File from the AutoEQ homepage. Just as Oratory, AutoEQ also cannot know where my personal resonance peak is.
Careful. You might get blasted here for using a "flawed methodology" :rolleyes:
https://www.reddit.com/r/oratory1990/comments/fr9k1r/_/flydswi
 

isostasy

Senior Member
Joined
Oct 19, 2022
Messages
354
Likes
628

I think we're all talking at cross-paths here. I'm fairly confident @Chromatischism is talking about the resonance peak you hear with a specific IEM (in this case the ER4XR), and is fully aware this is not going to be the same with all IEMs, nor will it necessarily correlate perfectly with the precise resonant frequency of your ear canal. Of course the quote you've shared is correct for finding the exact resonant frequency of the ear canal, but this isn't what we're concerned with when listening to certain IEMs, especially Etymotics, because the insertion depth will vary where this occurs due to the effective shortening of the canal. In this case it's perfectly reasonable to use a sine swipe in the general area we're concerned with to find where the peak resides. I'd suggest software that lets you control the frequency rather than simply a recorded sine sweep, as that way you'll be able to more precisely determine where the peak is. I found a nice open source app on Android simply called 'signal generator' which works perfectly, for example.

er4xr_insertion.png

Credits Earfonia, sorry I can't find the original link but I believe it's here on ASR somewhere.
 

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,348
Likes
1,804
Of course the quote you've shared is correct for finding the exact resonant frequency of the ear canal, but this isn't what we're concerned with when listening to certain IEMs
Yes it is. An acoustic cavity blocked at both ends (in this case by the ear drum on one end and the IEM on the other) will form a half-wavelength resonator with the resonance frequency dependent on the length of the cavity (here affected by IEM insertion depth).
In this case it's perfectly reasonable to use a sine swipe in the general area we're concerned with to find where the peak resides.
No it isn't. For the reasons Oratory provides. And there's a very good reason he says to use pink noise instead, and that's because it (along with music tracks that also have a similarly smooth broadband spectrum) have been scientifically determined in blind tests to produce the most discriminating and reliable listener judgements, as I've mentioned previously.
 

isostasy

Senior Member
Joined
Oct 19, 2022
Messages
354
Likes
628
Yes it is. An acoustic cavity blocked at both ends (in this case by the ear drum on one end and the IEM on the other) will form a half-wavelength resonator with the resonance frequency dependent on the length of the cavity (here affected by IEM insertion depth).

No it isn't. For the reasons Oratory provides. And there's a very good reason he says to use pink noise instead, and that's because it (along with music tracks that also have a similarly smooth broadband spectrum) have been scientifically determined in blind tests to produce the most discriminating and reliable listener judgements, as I've mentioned previously.

I still think we're not talking about the same thing here. Of course I agree with you that playing a sine sweep all the way from 0 to 20kHz isn't a good listening test, but I, @Chromatischism , and @Dazerdoreal were not speaking about that, we were talking about the resonant peak in a very narrow band. You can't seriously be trying to argue that playing a sine wave up and down this range isn't going to reveal to you where this peak is? Look at the graph above, it's at least a good 15dB, it's really obvious. I agree, you're not going to be able to discover the precise centre of the peak, nor the Q factor, by listening, but you can't tell @Dazerdoreal they're wrong to have found a resonant peak at 9kHz by using a sine sweep.

I would like to see the insert microphones you use to measure the resonant frequencies of the IEMs you own in your ears. It is specialist equipment which @Dazerdoreal and the rest of us are not lucky enough to own, so sine sweeps are a compromise to demonstrate roughly where the resonant peaks are for us.

n.b. can you please provide the link to the reddit thread you've taken the screenshot from? It would be helpful to have the context.
 
Last edited:
OP
D

Dazerdoreal

Active Member
Joined
Oct 5, 2022
Messages
212
Likes
216
Thank you for your comment. I can see why his method is even more precise if you want 100% perfect results, but I am sure ours is "good enough".
But I am aware that this forum rather targets the perfectionists, so I can see why someone is not happy with a "just good" solution.

I can say for sure though that my EQ right now is way better than Oratorys recommended wide filter EQ settings which did not take the peak into account (again, this is not his fault). The position I found is at least almost correct for sure. The slight error margin is not a big deal to me, nothing sounds wrong in that area anymore. :)

 

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,348
Likes
1,804
I still think we're not talking about the same thing here. Of course I agree with you that playing a sine sweep all the way from 0 to 20kHz isn't a good listening test, but I, @Chromatischism , and @Dazerdoreal were not speaking about that, we were talking about the resonant peak in a very narrow band. You can't seriously be trying to argue that playing a sine wave up and down this range isn't going to reveal to you where this peak is? Look at the graph above, it's at least a good 15dB, it's really obvious. I agree, you're not going to be able to discover the precise centre of the peak, nor the Q factor, by listening, but you can't tell @Dazerdoreal they're wrong to have found a resonant peak at 9kHz by using a sine sweep.

I would like to see the insert microphones you use to measure the resonant frequencies of the IEMs you own in your ears. It is specialist equipment which @Dazerdoreal and the rest of us are not lucky enough to own, so sine sweeps are a compromise to demonstrate roughly where the resonant peaks are for us.

n.b. can you please provide the link to the reddit thread you've taken the screenshot from? It would be helpful to have the context.
It's not a screenshot, it's a link (just click on it). Looks like you haven't fully read what Oratory said, so I'll quote the relevant part below. He's not saying to use insert mics, he says to do this:
using a broadband noise (pink noise or IEC noise) and a filter - set the filter to positive gain (+6 dB or so) and a rather narrow band (Q=4-6). Then enter different frequencies for the filter, and try to find out which one sounds worst, meaning at which frequency the shrieking sound produced by the filter sounds the worst - then apply negative gain at that frequency. It's of course not very precise, but better than listening to sine sweeps and getting false results.

And before you say this only applies to over-ear headphones after seeing the context, he says the same for IEMs, the principle is the same:
Would there be any point in trying to equalize that region if the ear canal resonance has a lot of variation?
If you know the frequency of the ear canal resonance in your ear you can try to compensate it with a corresponding notch. It's not something you have to do though. Frequency and amplitude of that resonance vary enormously from person to person.
If you want to experiment, play pink noise and use a narrow band filter (Q ~6) with positive gain of say 6 dB. Vary the frequency from 6 to 9 kHz and see at which frequency it sounds the most disturbing, then reduce the gain until that "disturbingness" disappears. Make sure to double check, listen to music you know and see whether that actually makes things sound more natural or not.
 

MayaTlab

Addicted to Fun and Learning
Joined
Aug 15, 2020
Messages
924
Likes
1,512
I still think we're not talking about the same thing here. Of course I agree with you that playing a sine sweep all the way from 0 to 20kHz isn't a good listening test, but I, @Chromatischism , and @Dazerdoreal were not speaking about that, we were talking about the resonant peak in a very narrow band. You can't seriously be trying to argue that playing a sine wave up and down this range isn't going to reveal to you where this peak is? Look at the graph above, it's at least a good 15dB, it's really obvious. I agree, you're not going to be able to discover the precise centre of the peak, nor the Q factor, by listening, but you can't tell @Dazerdoreal they're wrong to have found a resonant peak at 9kHz by using a sine sweep.

If you have a pair of reasonably well setup speakers, maybe even headphones, you can apply a fairly strong, narrow peak via EQ, neither too low or high in the spectrum, in a range that measures well, and sounds smooth when manually sweeping through it, to assess on your own whether or not you can recognise where the peak is with that combination of Q / magnitude. You can even ask someone else to do it and try to see whether or not you can successfully locate the peak.

Given the sort of Q / magnitude that most headphones can exhibit in the treble it should be fairly easy peasy to localise quite a few of the "peaks" in question. Now the question rather becomes "is what I heard a "peak", or a normal elevation surrounded by two dips ?", and which Q / magnitude is it - which I don't think as you said is easy to determine by manually sweeping through a part of the spectrum.

I mean, it should be fairly easy to notice this sort of "peak" (which in the case of the Clear MG rather is a "dip" surrounded by bigger "dips") when manually sweeping through that area :
Screenshot 2022-11-27 at 18.22.46.png


And you can do the test Oratory suggests to perform and see for yourself which one you think works best for you in terms of locating "peaks".

Another possibly more complicated question is whether or not you should be hearing certains peaks / dips in the treble or not (I.e such FR would be preferred to you or not), but in any case it's quite logical to suggest that IEMs' resonant peaks are unlikely to be desirable for most people unless they happen to exactly align with where you should have one, given their origin.


I would like to see the insert microphones you use to measure the resonant frequencies of the IEMs you own in your ears. It is specialist equipment which @Dazerdoreal and the rest of us are not lucky enough to own, so sine sweeps are a compromise to demonstrate roughly where the resonant peaks are for us.

Oratory was responding to a question regarding measuring a pair of over-ears. For these you can use in-ear mics indeed, but above a certain frequency they can become quite misleading in terms of the notion of "peaks" or "dips" depending on their design. Ideally you'd use a probe tube mic near the eardrum. If you use ear canal entrance mics (either blocked or open), you'll need to apply a transfer fonction to get what the measurement at the eardrum would look like, and above 7-8kHz they may already struggle to accurately describe the relative differences between headphones anyway, let alone the absolute values. And if you use anything that's past the canal entrance, the frequency at which errors start to occur is likely to be lower : https://audiosciencereview.com/foru...ted-harman-oe-curve-at-home.28130/post-975888

It's the best method to go beyond EQing headphones after ear simulator measurements or by ear, but it isn't completely straightforward either.

For in-ears it's a lot more difficult if the pair of headphones are heavily reliant on seal to perform well as the wire or the probe's tube can compromise it, and anything that isn't near the eardrum will introduce errors quite low in the spectrum.

Best implementations either locate a MEMS mic in the bore of the IEM, but then your measurements will only be valid up to several hundred Hz, perhaps 1kHz or so (here what Harman did for their in-ear articles) :
Screenshot 2022-11-27 at 18.18.07.png


Or pass the probe tube through a dedicated hole (occasionally seen in real ear measurements for hearing aids) :
Screenshot 2022-11-27 at 18.08.36.png


A cruder approach that I've occasionally tried at home, sacrificing an ear tip :
tempImageYlj3Uw.png

For in-ears with a decent feedback mechanism any leak with that approach should be minimal enough to be corrected by the feedback mechanism, but I'd still be a little hesitant to use it for passive IEMs.

And there's a very good reason he says to use pink noise instead, and that's because it (along with music tracks that also have a similarly smooth broadband spectrum) have been scientifically determined in blind tests to produce the most discriminating and reliable listener judgements, as I've mentioned previously.

The format of the test Oratory suggests to perform isn't similar to the one Harman performed at several occasion, the latter's research is unrelated to assessing whether or not Oratory's method is useful or not to locate peaks. *sigh*
 

MacClintock

Senior Member
Joined
May 24, 2023
Messages
435
Likes
769
"Dr. Olive and his researchers conducted blind studies on 283 individuals in four countries and 11 test locations. Although most of the participants were Harman employees, the science team was careful to achieve diversity in their gender, age, and listening experience."

You may or may not like what the average (?) listener likes.

I don't worry about it.
This quotation must be from a different paper, because for the In-Ear target (2017, smoothed version 2019), they used only about 20 people.
 

MacClintock

Senior Member
Joined
May 24, 2023
Messages
435
Likes
769
AutoEQ uses a modified version of the Harman target with 2dB less deep bass, for both headphones and IEMs.
I have suspected this. Do you have the exact source for this? And do you know the motivation? Does this modified target also enter in the compliance score point calculation?
 

IAtaman

Major Contributor
Forum Donor
Joined
Mar 29, 2021
Messages
2,273
Likes
3,836
I do agree with the subject statement that In-Ear target sounds off. Over-ear target works fine for me - all my headphones sound as good as they can with a tuning EQ'ed to harman target. I can not say the same for IE target however. I pretty much bought all the "hyped" cheap IEMs in an effort to find one that I like and can wear for a relatively longer duration, and when EQed to IE targets they all sound off and tire me out after a short while. All of them. It could as easily be me, but I would like to blame the target.

@Chromatischism , have been using your Truthear Zero EQ for about an hour now with some added bass (70hz +4db Lowshelf) and so far so good.
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
7,540
Likes
12,011
I have suspected this. Do you have the exact source for this?
The .csv in every AutoEQ preset contains the target in raw, numerical form: https://github.com/jaakkopasanen/Au...ts/oratory1990/over-ear/AKG K371/AKG K371.csv

If you extract it and overlay it onto stock Harman, you get:
AutoEQ Target vs Harman OE2018.png

Same story for IEMs.

And do you know the motivation?
Jaakko's motivation was that most headphones have strong bass roll-off, and so to reduce the amount of gain and pre gain required to get them to the target, Jaakko modified it to be easier to meet.

While IEMs typically don't have that issue, it was done there as well. I'm guessing for consistency's sake.

The decision to modify Harman for this reason has been a controversial one from the beginning.

Does this modified target also enter in the compliance score point calculation?
That I don't know. Perhaps @jaakkopasanen can give you an answer.
 
Top Bottom