• 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!

Best IEMs identified in Harman (Olive) research paper

preload

Major Contributor
Forum Donor
Joined
May 19, 2020
Messages
1,554
Likes
1,701
Location
California
Might be a long shot, but does anyone know the identity of the two IEMs circled in red below:
A Statistical Model That Predicts Listeners’ Preference Ratings of In-Ear Headphones: Part 2 – Development and Validation of the Model
1634341118940.png


Edit: these were the IEMs tested in the part 1 of the paper
1634410812037.png
 
Last edited:

enricoclaudio

Major Contributor
Audio Company
Joined
Jan 7, 2021
Messages
1,106
Likes
2,157
Location
Houston, TX - USA
That would be a question for @Sean Olive to answer.
 

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,348
Likes
1,804
I reckon the highest scoring one is the AKG N20. Why? In the paper, the frequency response of five of the highest scoring IEMs is shown:

Screenshot_20211016-035719_Acrobat for Samsung.png


To my eyes, it looks like the green curve has the lowest average deviation from the Harman target (black), and the flattest overall tonal tilt, so would be the highest scoring. From the plot you posted above the highest scoring model looks to have a predicted rating of just below 70, maybe 68-69. Looking at AutoEQ's ranking table, the AKG N20 looks like a good candidate, with a predicted rating of 68, fairly flat overal slope (-0.21, indicating a slightly warm tilt, which fits with the above green curve's bass hump centered around 200 Hz). Plus it's a well-known brand (owned by Harman no less), and was released before the paper (2017). Now comparing Oratory's measurements of the N20 to that green curve, I think we have a match (note the three peaks at 2.75, 6 and 9 kHz in both):

Screenshot_20211016-042348_Acrobat for Samsung.png


I can't be bothered to sleuth the other one out, but it looks like it's probably the red curve in the first graph I posted, and has a predicted rating of ~62-63.

Edit: just realised all the models tested are named in this presentation (the AKG N20 being one of them):

Screenshot_20211016-044217_Acrobat for Samsung.png


So that could help to identify the others, by cross referencing them with the ratings on AutoEQ's ranking and Oratory's (or Crinacle's) FR measurements.
 
Last edited:

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,348
Likes
1,804
That list contains very outdated IEM models which don’t represent the actual market, to be honest…

Sure, I was just trying to answer the OP's actual question, which was about the identity of the models tested in the research paper.
 

Merkurio

Senior Member
Joined
Jul 18, 2018
Messages
301
Likes
512
Get a Moondrop Kato. Infinitely better than these.

Better than Illumination? :p

BTW, you won’t reply my question about your thoughts comparing the Kato to the ER4SR!

7Hz Timless also gained a lot of hype recently, FR wise is very similar to Moondrop VDSF.
 

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,348
Likes
1,804
They both have higher ( and currently highest) preference rating than that old paper

Once again, the OP was not asking for IEMs with the highest predicted preference rating, which is actually (just) the AKG N400 with a score of 89 (and flatter, more neutral overall tonal slope of just -0.11), according to AutoEQ's ranking. Instead they were asking for the identity of the two highest scoring IEMs from the research paper, I suspect because these were the two that received the highest actual average scores from listeners in double blind tests. Remember that all these other calculated ratings are predicted scores (albeit based on a model that had a very high 91% correlation with the actual ratings from the study).
 

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,645
Likes
2,813
Once again, the OP was not asking for IEMs with the highest predicted preference rating, which is actually (just) the AKG N400 with a score of 89 (and flatter, more neutral overall tonal slope of just -0.11), according to AutoEQ's ranking. Instead they were asking for the identity of the two highest scoring IEMs from the research paper, I suspect because these were the two that received the highest actual average scores from listeners in double blind tests. Remember that all these other calculated ratings are predicted scores (albeit based on a model that had a very high 91% correlation with the actual ratings from the study).

Yes agreed, I saw your later reply to someone else.

I don't think sharing this JBL and Samsung highest predicted score was a complete waste to the OP though, who "liked" the replies.

I think it's at least interesting know which models have highest predicted score.
 

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,645
Likes
2,813
Get a Moondrop Kato. Infinitely better than these.

Better than the JBL Club Pro+ shown above? How?

The real killer of the JBL is the app's EQ - best EQ app of all IEMs by far.

So it starts with closest Harman curve of all IEMs measured so far - and then you can really adjust to personal preference with the App. But at least you are EQ'ing from a known standard.

And the EQ is stored in the earbuds , for all sources.

1634437632302.png
 

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,348
Likes
1,804
Anyway, back on topic, looking further at Sean Olive's presentation I linked to earlier, I now believe the two 'IEMs' circled in the OP are actually the Harman Target (highest scoring), and a modified 'Harman Target 2' (second highest scoring). After scaling this puts the Harman target at ~100, which is standard practice for Harman (and used by Oratory and AutoEQ in their predicted rating calculations):

Screenshot_20211017-013607_Acrobat for Samsung.png


So ignoring the targets, the highest actual rating given to an IEM in the study was just above 80 (after scaling). Comparing with the plot below and the list of models tested (with prices), this looks like it's in fact the AKG K3003 (Reference Filter):

Screenshot_20211017-013620_Acrobat for Samsung.png


The second highest scoring IEM from the study isn't far behind though with an average rating given of just below 80, for almost a tenth of the cost, and looking at the price (seems to be $130 to me), and frequency response as I previously noted, I think this is the AKG N20.
 
OP
preload

preload

Major Contributor
Forum Donor
Joined
May 19, 2020
Messages
1,554
Likes
1,701
Location
California
Anyway, back on topic, looking further at Sean Olive's presentation I linked to earlier, I now believe the two 'IEMs' circled in the OP are actually the Harman Target (highest scoring), and a modified 'Harman Target 2' (second highest scoring). After scaling this puts the Harman target at ~100, which is standard practice for Harman (and used by Oratory and AutoEQ in their predicted rating calculations):

View attachment 159526

So ignoring the targets, the highest actual rating given to an IEM in the study was just above 80 (after scaling). Comparing with the plot below and the list of models tested (with prices), this looks like it's in fact the AKG K3003 (Reference Filter):

View attachment 159528

The second highest scoring IEM from the study isn't far behind though with an average rating given of just below 80, for almost a tenth of the cost, and looking at the price (seems to be $130 to me), and frequency response as I previously noted, I think this is the AKG N20.

So it looks like #1 and #2 were <$125. #3 looks like it was $1,000, which corresponds to the AKG 3003. Or am I reading it wrong?

It looks like I'm your two charts there are 4 distinct iem's with a pref score >75 that we should be able to match up?

Thank you btw.
 
Last edited:

Jimbob54

Grand Contributor
Forum Donor
Joined
Oct 25, 2019
Messages
11,067
Likes
14,700
So it looks like #1 and #2 were <$125. #3 looks like it was $1,000, which corresponds to the AKG 3003. Or am I reading it wrong?

It looks like I'm your two charts there are 4 distinct iem's with a pref score >75 that we should be able to match up?

Thank you btw.
I think GaryH is saying the k3003 is the highest circled iem in your OP and the N20 is second.

It looks to me that if you are in the market for Harman matching iem and can't find measurements for the current crop of models, picking a wireless model from one of the Harman /Samsung stable of brands might be best.

That said I have the galaxy buds plus and they sound like ass to me.
 

JustAnandaDourEyedDude

Addicted to Fun and Learning
Joined
Apr 29, 2020
Messages
518
Likes
819
Location
USA
So it looks like #1 and #2 were <$125. #3 looks like it was $1,000, which corresponds to the AKG 3003. Or am I reading it wrong?

It looks like I'm your two charts there are 4 distinct iem's with a pref score >75 that we should be able to match up?

Thank you btw.
I think GaryH is saying the k3003 is the highest circled iem in your OP and the N20 is second.
There are 32 dots on each graph, while 30 tested earphone models are listed in the OP and in Post #3. It seems to me that GaryH is saying that the circled IEMs in the OP, which have the highest actual preference score, are 'Harman Target' and 'Harman Target 2' (maybe EQ'd IEMs? I have not read the paper). It appears that both actual and predicted scores have been scaled so that the 'Harman Target' scores 100, and both the Targets cost a made-up $100. It also seems GaryH is saying that top two actual preference scorers with their native FR without EQ (which preload refers to as #3 and #4) correspond to the AKG K3003 at $1000 and AKG N20 at $130.
 
Last edited:
Top Bottom