Harman curve for loudspeakers

[Phorize]

I’ve been pondering the harman curve this morning, or more specifically why the convention is to measure headphones with references to the harman curve, but to measure loudspeakers with reference to an ideal flat response. Perhaps for professional monitoring a flat loudspeaker response is ideal, but if Harman found that users prefer a non linear frequency response in a headphone, is there any research that suggests a similar preference for non linear frequency response in speakers and if so why is that not the reference for non professional loudspeakers?

 

[Ilkless]

I’ve been pondering the harman curve this morning, or more specifically why the convention is to measure headphones with references to the harman curve, but to measure loudspeakers with reference to an ideal flat response. Perhaps for professional monitoring a flat loudspeaker response is ideal, but if Harman found that users prefer a non linear frequency response in a headphone, is there any research that suggests a similar preference for non linear frequency response in speakers and if so why is that not the reference for non professional loudspeakers?

The Harman headphone curve is the response of a flat loudspeaker (Salon2 IIRC) in a moderately reflective room (intended to be a representation of domestic listening), measured by a calibrated head-and-torso simulator that acts as a surrogate for a human listener.

 

[dasdoing]

https://www.aes.org/e-lib/browse.cfm?elib=17839

 

[dasdoing]

Perhaps for professional monitoring a flat loudspeaker response is ideal

my research sugests that many engeniers use a steady roll-off after 1k, too (without the bass boost), but the steepness can't be meassured because it depends on distance and how reflective the room is.
actualy this is a gap in research imo. there must be a method to actualy meassure the real brightness if you include the time domain.
the only try to do so I found is this from @jlo : http://www.ohl.to/calculators/targetcurve.php; though experience seam to suggest F2 point beeing at 1000Hz

 

[dasdoing]

I am using Bob Katz's method atm:

https://www.facebook.com/bobkatz24bit/posts/1416323988391278

which summerizes to "try out various steepnesses after 1000Hz after none of your music sound too bright or too dull"

 

[andreasmaaan]

I’ve been pondering the harman curve this morning, or more specifically why the convention is to measure headphones with references to the harman curve, but to measure loudspeakers with reference to an ideal flat response. Perhaps for professional monitoring a flat loudspeaker response is ideal, but if Harman found that users prefer a non linear frequency response in a headphone, is there any research that suggests a similar preference for non linear frequency response in speakers and if so why is that not the reference for non professional loudspeakers?

According to the Harman targets, loudspeakers are supposed to measure flat anechoically (in a reflection-free environment).

When you place a conventional loudspeaker in a well set-up listening room, the combination of direct sound (which should be flat) and reflections will combine to produce a downward sloping response (similar to the Harman target curve for headphones).

And as @Ilkless mentioned, the Harman target curve for headphones was derived from the in-room response curve produced by a flat-measuring speaker (Revel F208 IIRC) in a well-treated room.

 

[thewas]

The Harman headphone curve is the response of a flat loudspeaker (Salon2 IIRC) in a moderately reflective room (intended to be a representation of domestic listening), measured by a calibrated head-and-torso simulator that acts as a surrogate for a human listener.

I also thought and hoped so in the beginning, but it isn't, it was derived by EQing the Revels first to measure flat at the listeners position (mistake #1) and then by listeners adjusting bass and treble to their liking (mistake #2) by listening to some music (mistake #3 - audios circle of confusion).
In the end fortunately by the adjustments the average listeners chose it luckily doesn't measure measure very different than the Revels without any EQ except the lower bass.

 

[dasdoing]

When you place a conventional loudspeaker in a well set-up listening room, the combination of direct sound (which should be flat) and reflections will combine to produce a downward sloping response (similar to the Harman target curve for headphones).

the problem with the Harman curve is that pros don't balance the music in a "well set-up listening room" but a well treated one. also the precieved brightness will not translate if your listening room doesn't match the Harman room

 

[andreasmaaan]

the problem with the Harman curve is that pros don't balance the music in a "well set-up listening room" but a well treated one. also the precieved brightness will not translate if your listening room doesn't match the Harman room

Not sure I understand the difference between “well set-up” and “well-treated”?

 

[dasdoing]

Not sure I understand the difference between “well set-up” and “well-treated”?

the later will have no bass boom and treated first reflection points which will alter the percieved brightness

my reasoning is the following: we shouldn't recreate the average well set-up listening room, but the average mastering engenier's room; because this is what the music has been balanced to. the problem is that (HF) curves don't seam to translate between rooms

 

[andreasmaaan]

the later will have no bass boom and treated first reflection points which will alter the percieved brightness

Treated with absorption?

my reasoning is the following: we shouldn't recreate the average well set-up listening room, but the average mastering engenier's room; because this is what the music has been balanced to. the problem is that (HF) curves don't seam to translate between rooms

That makes sense, although I would expect the steady-state response to become brighter in the high frequencies as more absorption of early reflections was added, not more downward-sloping.

 

[dasdoing]

meassurements of 2 WDR (huge German broadcasting institution) control rooms (Schoeps mk2 mics; pretty flat).

prety bad, right? lol
but we can average those to being flatish until 1500ish and then tilting downwards

 

[dasdoing]

Treated with absorption?

.

yea

That makes sense, although I would expect the steady-state response to become brighter in the high frequencies as more absorption of early reflections was added, not more downward-sloping.

you take them away from the room response, you take them away form the overall response (direct+room). that is why flat target sounds too bright, because this never happend in treated and non-equed a pro room

 

[dasdoing]

this seams to confirm the 1000Hz for the (varying) tilt point https://www.audiosciencereview.com/...tudio-monitor-review.14719/page-4#post-459114

 

[thewas]

meassurements of 2 WDR (huge German broadcasting institution) control rooms (Schoeps mk2 mics; pretty flat).

View attachment 101598

prety bad, right? lol
but we can average those to being flatish until 1500ish and then tilting downwards

Do you know which monitors were used there?

 

[andreasmaaan]

Ok, I’m just having trouble following you because you say:

the problem with the Harman curve is that pros don't balance the music in a "well set-up listening room" but a well treated one. also the precieved brightness will not translate if your listening room doesn't match the Harman room

And then:

my reasoning is the following: we shouldn't recreate the average well set-up listening room, but the average mastering engenier's room; because this is what the music has been balanced to. the problem is that (HF) curves don't seam to translate between rooms

But, due to heavier absorption, the average mastering engineer’s room will have a brighter steady-state response than Harman’s.

FWIW, my view is that looking at the in-room response is the wrong way to go in the first place.

A speaker should measure anechoically flat with smooth dispersion. How it should measure in a particular room should be how an anechoically flat speaker with smooth dispersion measures in that room.

 

[thewas]

this seams to confirm the 1000Hz for the (varying) tilt point https://www.audiosciencereview.com/...tudio-monitor-review.14719/page-4#post-459114

The good old Bob Katz target which I also use currently, flat till 1 kHz and then usually dropping around 6dB till 20 kHz.
https://www.audiosciencereview.com/...ike-to-see-reviewed.10943/page-25#post-359319
Mind you it must suit the directivity of the loudspeakers and reverberation of the room, as it can be seen from the Dynaudio example.

 

[Music1969]

I tried various downward sloping cruves.

But I seem to prefer flat to 20kHz - except a 3-6 dB bass bump (depending on old and new recordings) from 20Hz to 200Hz.

I can select the 2 different bass profiles quite easily using convolution engine.

 

[dasdoing]

Do you know which monitors were used there?

this is the dataset http://audiogroup.web.th-koeln.de/wdr_irc.html

if found these pictures:

CR1:

this picture of a meassurement sugest the actual set beeing used (note that the meassurements I showed are not part of the BRIR meassurements)

CR7 seams to have only one set:

 

[thewas]

this is the dataset http://audiogroup.web.th-koeln.de/wdr_irc.html

if found these pictures:

CR1:

View attachment 101602

this picture of a meassurement sugest the actual set beeing used (note that the meassurements I showed are not part of the BRIR meassurements)

View attachment 101603

CR7 seams to have only one set:

View attachment 101604

Thank you, CR7 measurement confirms well as it shows the typical B&W presence dip.