• 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!

Headphones and the Harman target curve

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,205
Location
North-East
I'm confused. Maybe you can clarify this.

You said this measurement was based on microphones in your ears. Are the microphones at the entrance of the ear canal (ie blocked meatus) or probe mics at your ear drums (DRP or ear drum reference point)

If they are blocked meatus measurements which they appear to be then you are not including the 3 kHz ear canal resonance.

The Harman Target Curve is measured at the end of the ear canal and includes the 3 khz resonance. So have you accounted for that when equalizing to the Harman Target? If you are equalizing a blocked meatus measurement to a curve based on a DRP that would explain why it sounds bad. Just checking.

My measurements with in-ear mics were blocked-meatus. But the Harman target corrections I tested against were using EQ settings reported by others, from Oratory1990 to now, Amir. The corrections that sound best to me are measured with in-ear mics with a mostly flat response up to about 4kHz, or about as far as I trust my measurements to be consistent, and an elevated bass shelf of about 2-3dB.

The EQ settings to Harman target as reported by others sound too bright and too bass-shy to me (possibly the same thing). The goal for me was to find what sounds good to me, it wasn't a research into a "better" curve. My assumption was that there may be sufficient variations between my ears and others that it would be worth the effort to try to find what works best for me, and so far, it's not what others seem to prefer.
 

Sean Olive

Senior Member
Audio Luminary
Technical Expert
Joined
Jul 31, 2019
Messages
329
Likes
3,004
My measurements with in-ear mics were blocked-meatus. But the Harman target corrections I tested against were using EQ settings reported by others, from Oratory1990 to now, Amir. The corrections that sound best to me are measured with in-ear mics with a mostly flat response up to about 4kHz, or about as far as I trust my measurements to be consistent, and an elevated bass shelf of about 2-3dB.

The EQ settings to Harman target as reported by others sound too bright and too bass-shy to me (possibly the same thing). The goal for me was to find what sounds good to me, it wasn't a research into a "better" curve. My assumption was that there may be sufficient variations between my ears and others that it would be worth the effort to try to find what works best for me, and so far, it's not what others seem to prefer.

OK, so you simply applied the Harman Target EQ settings to the headphone based on measurements of others, not based on your measurement. When I saw something that looked like the Harman curve superimposed on your blocked meatus measurements it confused me. Thanks!
 

Yorkshire Mouth

Major Contributor
Joined
Apr 22, 2020
Messages
1,254
Likes
1,154
Location
God's County - Yorkshire
Okay, a few issues with and questions about the methodology. Please feel free to chip in.

1 - ‘Good speakers in a good room’. That’s okay, I can see the point. But why not go the whole hog? Why not go to Abbey Road and use the mixing room there? Whatever the faults and flaws, our target must surely be to hear the music exactly as it was heard when mixed.

2 - Play this back into the same type of measuring equipment used to measure the headphones. Great idea, can’t argue with that.

3 - Okay, but why not just stop there. Take that measurement, and that’s your curve. By the way, have Harman ever released this curve? I’d have thought it should be in the data somewhere. Anyone?

4 - Get trained listeners to listen to that (and other curves), and come up with an average of what they prefer. What? Why bring preference into it at all. Surely we should be asking for accuracy. The legitimate next scientific step here is to sit those trained listeners in the ‘good room with good speakers’ (or Abbey Road, see above), and with headphones tuned to the curve measured earlier, then play a variety of pieces of music both through the speakers and headphones. And ask the trained listeners if they sound the same. Not if they prefer that, or another sound. Just check that the measuring equipment is doing what it’s supposed to do. If not, tweak the curve.

5 - A few years later get untrained listeners involved. At this point, I’m out. Why on Earth is this either necessary or desirable, in regards to what we’re discussing?

There is some validity in what Harman have done, to a point. But this is supposed to be science. The kit should be there for one reason and one reason only, to faithfully and accurately reproduce the music, in as far as that’s possible. And that’s what the curve should be.

The annoying thing is, we have the ability to do just this. Harman have come pretty close to doing just this. The whole ‘preference’ question either shouldn’t be there at all, or just a minor footnote.

By the way, I suspect the end result wouldn’t be a million miles away from Harman.
 

o7_brother

Active Member
Joined
Oct 8, 2019
Messages
106
Likes
326
Location
Portugal
By the way, I suspect the end result wouldn’t be a million miles away from Harman

To sum up your various points, I think you've answered your own question here.

I recall reading that when the listeners were asked to rank something based on "neutrality", you get similar results as if you had asked what they preferred.

As it turns out, we tend to equate these things in our minds; we prefer what sounds "right" to us, with no audible flaws that would stand out, which can be interpreted as "neutral".
 

Yorkshire Mouth

Major Contributor
Joined
Apr 22, 2020
Messages
1,254
Likes
1,154
Location
God's County - Yorkshire
I do find it interesting that, as soon as ‘the great unwashed’ become involved, they want more bass.

The fact that people not committed to accurate reproduction want to turn the bass up is nothing new.

Every DAC/amp test Amir does, he looks for linearity of frequency. If the line wasn’t flat, but increased bass, and Amir said “That’s a good thing because people prefer more bass”, what would our response be? I suspect the whole raison d'être of the group would be undermined.

How is following the herd unscientific with a DAC, but great with headphones?

This shouldn’t be about preference, just accuracy.

If you get accurate kit, then want to turn the bass up, you can.
 

Jimbob54

Grand Contributor
Forum Donor
Joined
Oct 25, 2019
Messages
11,066
Likes
14,698
I do find it interesting that, as soon as ‘the great unwashed’ become involved, they want more bass.

The fact that people not committed to accurate reproduction want to turn the bass up is nothing new.

Every DAC/amp test Amir does, he looks for linearity of frequency. If the line wasn’t flat, but increased bass, and Amir said “That’s a good thing because people prefer more bass”, what would our response be? I suspect the whole raison d'être of the group would be undermined.

How is following the herd unscientific with a DAC, but great with headphones?

This shouldn’t be about preference, just accuracy.

If you get accurate kit, then want to turn the bass up, you can.

I'm not convinced assessing a headphone as stock against any target, Harman or otherwise, is massively useful on its own. Fortunately, what Amir is really testing is the ability to successfully eq the headphone to that target or in that direction.

He knows he prefers Harman. Can he get the cans to something he likes and (because the science tells him so) by extension a significant proportion of the readership.

Im not sure we can ever assess absolute accuracy in headphones due to the variances between listeners. So even if the mastering engineer says "these are my approved headphones that best reflect what I hear through my monitors", still no guarantee I will hear it them the same.
 

Yorkshire Mouth

Major Contributor
Joined
Apr 22, 2020
Messages
1,254
Likes
1,154
Location
God's County - Yorkshire
I'm not convinced assessing a headphone as stock against any target, Harman or otherwise, is massively useful on its own. Fortunately, what Amir is really testing is the ability to successfully eq the headphone to that target or in that direction.

He knows he prefers Harman. Can he get the cans to something he likes and (because the science tells him so) by extension a significant proportion of the readership.

Im not sure we can ever assess absolute accuracy in headphones due to the variances between listeners. So even if the mastering engineer says "these are my approved headphones that best reflect what I hear through my monitors", still no guarantee I will hear it them the same.

I agree in part, but you're in danger of making the perfect the enemy of the good.

You're right, we're never going to get it spot on. But we can aim to get it as close to that as possible.

Take the example I gave. If someone said to me "These headphones follow a curve. We got a dozen trained listeners to listen whilst sat at the mixing desk in Abbey Road, listening to the same piece of music on both their speakers, and headphones which followed the same frequency response curve as these. And this as close as we could get to sitting in that chair."

I agree that wouldn't be perfect. I'm just saying it'd be more objective than Harman's 'preference' methodology.

We might not be able to rid ourselves of subjectivity completely. But our goal should be to reduce it as much as possible. We shouldn't be saying we can't be completely objective, so let's not bother at all.
 

RHO

Major Contributor
Joined
Nov 20, 2020
Messages
1,182
Likes
1,087
Location
Belgium
I don't think Harman was searching for a curve showing the most accurate reproduction, but for a curve that would help them sell more headphones.
Hence untrained listeners and preference etc.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
SO tile.jpg
 

Yorkshire Mouth

Major Contributor
Joined
Apr 22, 2020
Messages
1,254
Likes
1,154
Location
God's County - Yorkshire
I don't think Harman was searching for a curve showing the most accurate reproduction, but for a curve that would help them sell more headphones.
Hence untrained listeners and preference etc.

I've grinned a bit at that, but also gone "Grrrr!" a bit, too.

We're an objectivist forum.

I'm going to repeat my question, as I fear it's become lost.

When Harman did their 'great speakers in a great room' test, what was the result BEFORE they started asking people their preference?

Does that curve exist?
 

Jimbob54

Grand Contributor
Forum Donor
Joined
Oct 25, 2019
Messages
11,066
Likes
14,698
I've grinned a bit at that, but also gone "Grrrr!" a bit, too.

We're an objectivist forum.

I'm going to repeat my question, as I fear it's become lost.

When Harman did their 'great speakers in a great room' test, what was the result BEFORE they started asking people their preference?

Does that curve exist?

I see your point/ question.

I suspect the answer lies in the papers which show how they got to the original Harman OE headphone curve. Perhaps the good Dr @Sean Olive could enlighten us as to what the curve without any preference looked like initially

EDIT- I wouldnt be surprised if it wasnt pretty much the first OE published curve but with a lower bass shelf
 
Last edited:

RHO

Major Contributor
Joined
Nov 20, 2020
Messages
1,182
Likes
1,087
Location
Belgium
I've grinned a bit at that, but also gone "Grrrr!" a bit, too.

We're an objectivist forum.

I'm going to repeat my question, as I fear it's become lost.

When Harman did their 'great speakers in a great room' test, what was the result BEFORE they started asking people their preference?

Does that curve exist?
This forum did not pay for the research. We are not entitled to any of the data Harman does not want to reveal.
It would be nice if they would give us the data you are asking for. But wouldn't hold it against them if they didn't.
 

preload

Major Contributor
Forum Donor
Joined
May 19, 2020
Messages
1,554
Likes
1,701
Location
California
Okay, a few issues with and questions about the methodology. Please feel free to chip in.

1 - ‘Good speakers in a good room’. That’s okay, I can see the point. But why not go the whole hog? Why not go to Abbey Road and use the mixing room there? Whatever the faults and flaws, our target must surely be to hear the music exactly as it was heard when mixed.

For starters, that suggestion would be blasphemy, since Abbey Road features B&W speakers, which do not follow the ideal Harman loudspeaker curve nor Harman's principles related to directivity.

3 - Okay, but why not just stop there. Take that measurement, and that’s your curve. By the way, have Harman ever released this curve? I’d have thought it should be in the data somewhere. Anyone?

Yes. It's in their paper.

4 - Get trained listeners to listen to that (and other curves), and come up with an average of what they prefer. What? Why bring preference into it at all. Surely we should be asking for accuracy.

Nope. The outcome of interest is perceived sound quality. For a minority of listeners, perceived sound quality equates with perceived accuracy of reproduction. For most listeners, it's simply what makes the recording sound good to them. Unfortunately the reality is that not all music is a live recording, and it is common for music to get mixed and adjusted. So there's no way to know what "accurate" really is. This is why listener preference score seems to be the most elegant and logical way to measure transducer sound quality to date, IMO.
 

Zensō

Major Contributor
Joined
Mar 11, 2020
Messages
2,740
Likes
6,740
Location
California
1 - ‘Good speakers in a good room’. That’s okay, I can see the point. But why not go the whole hog? Why not go to Abbey Road and use the mixing room there? Whatever the faults and flaws, our target must surely be to hear the music exactly as it was heard when mixed.

So much of music is terribly mixed and mastered, I have to wonder if this is what we actually want. Audiophiles sometimes put mixing/mastering engineers on a pedestal, but the quality of their work varies enormously, and they’re also often mixing/mastering primarily for translation across a wide variety of systems from high to very low quality, not ultimate sound quality.
 

Yorkshire Mouth

Major Contributor
Joined
Apr 22, 2020
Messages
1,254
Likes
1,154
Location
God's County - Yorkshire
For starters, that suggestion would be blasphemy, since Abbey Road features B&W speakers, which do not follow the ideal Harman loudspeaker curve nor Harman's principles related to directivity.

The Harman Curve is, by definition, for headphones, in attempt to dimiish the issues caused by headphone listening being different to loudspeaker listening (I’m using necessary shorthand here).

Abbey Road mixes using B&W 800Ds, which are quite excellent.
 

Yorkshire Mouth

Major Contributor
Joined
Apr 22, 2020
Messages
1,254
Likes
1,154
Location
God's County - Yorkshire
So much of music is terribly mixed and mastered, I have to wonder if this is what we actually want. Audiophiles sometimes put mixing/mastering engineers on a pedestal, but the quality of their work varies enormously, and they’re also often mixing/mastering primarily for translation across a wide variety of systems from high to very low quality, not ultimate sound quality.

I don’t disagree in general. But you can’t negate ‘poor’ mixing by having speakers/loudspeakers which balance out these problems by ‘going the other way’.

All hi-fi can aim to do is faithfully and accurately reproduce the mix. After that, it’s down to you.
 
Top Bottom