• 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 Harman In-Room Target Curve - why is it represented as a thin line?

preload

Major Contributor
Forum Donor
Joined
May 19, 2020
Messages
1,570
Likes
1,735
Location
California
I typically see the Harman target curve represented as a thin line on an amplitude vs. freq chart, as if the goal were to achieve that target precisely.

But looking at a review article written by Toole entitled "The Measurement and Calibration of Sound Reproducing Systems" (JAES 2015;63:512-541), the preferred "target" is actually depicted as not one, but a series of target curves that account for differences in: a) listener preference (okay, "experience"), and b) room reflectivity. It's not a single thin line. In fact, for the predicted steady-state curve that represents how highly rated loudspeakers measure in a typical room, there is still a region of variability/uncertainty, represented by the shaded region below.

All in all, it doesn't make sense to me to boil all of this information down into a single thin curve, which is how I see the Harman target represented pretty much everywhere. Perhaps someone can enlighten me.

1611883562500.png
 
Last edited:
Simply said as it can't be boiled down to a single line as first of all depends on the directivity characteristics of the loudspeaker, reverberation characteristics of the room and listening distance.
Secondly above preference values were just a result of a single experiment with only 7 trained and 4 untrained listeners which even had a too large deviation between them to be grouped together.
Headphones don't have the first problem and for them much broader tests with 250 participants were done, see more below
https://www.audiosciencereview.com/...snt-like-this-curve.19668/page-10#post-652590
https://www.audiosciencereview.com/...snt-like-this-curve.19668/page-10#post-652775
https://www.audiosciencereview.com/...ut-room-curve-targets-room-eq-and-more.10950/
 
imo the study is outdated (besides never been relevant for treated rooms) because we now have methods to meassure (filter) the direct sound. note that it even mentions flat direct sound beeing the goal
 
Simply said as it can't be boiled down to a single line as first of all depends on the directivity characteristics of the loudspeaker, reverberation characteristics of the room and listening distance.

Thank you for confirming. That was also my interpretation. Does anyone understand it differently?
I guess my next question is, why is the concept of a single line target propagating everywhere in audio forums, on audio websites, etc?

Secondly above preference values were just a result of a single experiment with only 7 trained and 4 untrained listeners which even had a too large deviation between them to be grouped together.

In other scientific disciplines, I would have expected there to be error bars or other ways of conveying that variation, in the paper itself. But I'm guessing the paper was well within the standards of engineering journals. I agree that is a horribly small sample size to be able to come up with a target, UNLESS the preferences all converged precisely around a single line (but we don't know because there were no error bars!!). Maybe it's in the original-original paper.

Headphones don't have the first problem and for them much broader tests with 250 participants were done, see more below

I'd like to focus on loudspeakers here, since there's another dedicated thread for headphone curves.
 
Last edited:
I'd like to focus on loudspeakers here, since there's another dedicated thread for headphone curves
Yes, I mentioned them only to show the difference why for headphones a general target curve makes more sense.
 
why is the concept of a single line target propagating everywhere in audio forums, on audio websites, etc?

Because audiophiles want a single "best" community approved outcome to shoot for. Misguided for sure, but, as can be seen in the headphone PEQ obsession, they will use any number of filter combinations to desperately attempt to achieve someone else's arbitrary curved line.

I imagine the boss runs a whiteboard marker over the board and exclaims:

"This is the official company target line going forward!
Engineering: See how close you can get.
Marketing: Talk up this preference line as being gospel. I want it on brochures, packaging, websites and forums 24/7.
Research: Get some of those trained listeners, offer them some of our cheap speakers or free headphones as incentive, put them in a room, do some limited testing and find those extra votes. And get your secretaries to write up some whitepapers!
Get to it guys"
 
next question is, why is the concept of a single line target propagating everywhere in audio forums, on audio websites, etc?

Quite like with everything else: laymen don't want to know the guts and the differences. If you are asking that question, you'll probably way off from understanding how little people care.

And anyways, some target always makes sense, since the feature space is literally infinite and you'd have a hard time aligning to unknown target.
 
@restorer-john and @Feelas have probably hit the nail on the head.

As for why *that" particular curve, depending on one's level of cynicism either follow the science or the money. Or possibly both
 
I wrote a massive essay here and forgot that it's not on topic.

Let's all not forget, that most audiophools aren't really keen on pro-audio knowledge, okay? That explains everything.

I think that the curve is somewhere nearby, since trained listeners are biased by what they assess to be neutral and untrained by what they've heard until then. It all makes sense, actually.
 
I typically see the Harman target curve represented as a thin line on an amplitude vs. freq chart, as if the goal were to achieve that target precisely.

But looking at a review article written by Toole entitled "The Measurement and Calibration of Sound Reproducing Systems" (JAES 2015;63:512-541), the preferred "target" is actually depicted as not one, but a series of target curves that account for differences in: a) listener preference (okay, "experience"), and b) room reflectivity. It's not a single thin line. In fact, for the predicted steady-state curve that represents how highly rated loudspeakers measure in a typical room, there is still a region of variability/uncertainty, represented by the shaded region below.

All in all, it doesn't make sense to me to boil all of this information down into a single thin curve, which is how I see the Harman target represented pretty much everywhere. Perhaps someone can enlighten me.

View attachment 109098

Thanks a lot for posting this graph in high resolution - some time ago a posted a rough photo of it as I feared copyright issues.
This graph goes right to the heart of the room curve discussion.
The untrained one is what most people like, as most are „untrained“. These people, me included, love a strong bass and certainly no dip in the high frequency. The trained one does probably not want to admit liking bass (like a chef not wanting ketchup) and does not like a high frequency boost, probably finding it a bit vulgar.
Same problem happens during wine tasting when the sweet cheap one just tastes sooo good...
Now, the average of trained and untrained is made up to rescue the theoretical dashed curve.... Is this then an „average“ listener, eh half-trained but not enough?
The problem of theee Room curve could not be better illustrated with this graph. Btw, the untrained bass boost follows the FletcherHenderson loudness curve‘s delta 20-80db - untrained listeners too.
I slightly modify the untrained curve for me, it is a lot of bass boost for at 80db level otherwise - sounds great.
 
imo the study is outdated (besides never been relevant for treated rooms) because we now have methods to meassure (filter) the direct sound. note that it even mentions flat direct sound beeing the goal

I just saw this and not sure I'm completely following what you're trying to say.
The ability to measure "direct sound" (if you're referring to the anechoic on-axis response) predates this study. I also would interpret the graphs to refer to how a loudspeaker with "flat direct sound" measures when considering a typical reflective room. Even "treated" rooms will have some degree of reflections (otherwise they would be anechoic chambers).
 
As you can see all of them are identical (including the predicted in-room response) except for the sub-bass region.

The sub-bass region is highly variable, but also highly important in a response.

this is the take away message here. Not trained vs untrained.
 
As you can see all of them are identical (including the predicted in-room response) except for the sub-bass region.

The sub-bass region is highly variable, but also highly important in a response.

this is the take away message here. Not trained vs untrained.[/QUOTE]

I'm not sure you're looking at the same thing. Referring to the graph I posted originally, there are clearly different preferred FR curves for trained vs. untrained. There is also variability in how flat on-axis loudspeakers measure in different rooms depending on the room's reflection characteristics (represented by the shaded region).
 
Because audiophiles want a single "best" community approved outcome to shoot for. Misguided for sure, but, as can be seen in the headphone PEQ obsession, they will use any number of filter combinations to desperately attempt to achieve someone else's arbitrary curved line.

I imagine the boss runs a whiteboard marker over the board and exclaims:

"This is the official company target line going forward!
Engineering: See how close you can get.
Marketing: Talk up this preference line as being gospel. I want it on brochures, packaging, websites and forums 24/7.
Research: Get some of those trained listeners, offer them some of our cheap speakers or free headphones as incentive, put them in a room, do some limited testing and find those extra votes. And get your secretaries to write up some whitepapers!
Get to it guys"
Not really. The segment of consumers that cares about target compliance is tiny.
 
Sarcastic clowns should remember that B&K arrived at a similar conclusion without going the "democratic average" way.

let's not get emotional.

they are pretty diferent actualy
correction-plots-2.png


I just saw this and not sure I'm completely following what you're trying to say.
The ability to measure "direct sound" (if you're referring to the anechoic on-axis response) predates this study. I also would interpret the graphs to refer to how a loudspeaker with "flat direct sound" measures when considering a typical reflective room. Even "treated" rooms will have some degree of reflections (otherwise they would be anechoic chambers).


the problem with all taget curves is that they might be a good average for the room and speaker-wall-listening distance they were developed in but the translation to another room will vary drasticly. also they are keeping the room boost; something that "sounds right" since we grew up listening to music in rooms, but is not realy neutral.
check this topic: https://www.audiosciencereview.com/...n-with-james-jj-johnston-serge-smirnov.21716/
I am still owning answers in that topic. the topic went a little down in my priorities since I am pretty happy with the curve I ended up with
 
let's not get emotional.

they are pretty diferent actualy
I did use "similar", even if I was a bit moody in my post. Personally, I think that EQing anechoically flat speakers above the Schroeder frequency is useless, these curves just describe the outcome in some normalized conditions, not a target.
 
Back
Top Bottom