• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

What is your favorite house curve

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
4,693
Likes
3,117
#81
Just implemented this. Works great for music but may need some tuning for HT (This is the B&K curve in case you are wondering). While it looks steeper than it is, it is only about a 5dB drop from 50Hz to 20K Hz./ Depending on the high end of your speakers, that portion of the target may need diddling/adjustment. View attachment 13275
That looks conservative. What I often start with is 3 db per decade which is about .9 db per octave. Your curve looks to be 2 db per decade. Not saying you need to use 3db/decade, just a point of reference. I usually stay flat 200 hz and below then adjust that by ear. The best amount down there seems dependent upon the room, how low the speaker in use goes and some amount of individual taste. I do usually end up with something like you are showing. A gentle hump 50-80 hz and shelving at lower frequencies.
 
Joined
Mar 17, 2016
Messages
46
Likes
19
#82
That looks conservative.
It is conservative. Given how easy it is to create new targets and implement them, I gave it a go but I had no expectations one way or the other how it would sound. I'm still evaluating but so far, so good. And you are right, the success (or lack of it) is a function of room size, treatment, speakers, listening distance and personal preference.
 

Wombat

Major Contributor
Joined
Nov 5, 2017
Messages
3,123
Likes
1,371
Location
Australia
#83
Here is the target curve (in dashed lines) for JBL Synthesis Arcos system for the main speakers. Looks like a 5 dB drop from 80 Hz to 20 kHz:

View attachment 10873

The faint gray is pre-correct, the blue, post correction.
Is the shelf boost for baffle-step compensation at the LF end?
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
489
#84
I think the B&K curve is probably good enough. It is close to what many more modern estimates, like Olive's and others, consider a good curve, although the modern curves tend to have an increasingly negative slope above about 15k. Not a big deal, though. But, sometimes CD playback and its potential filter artifacts benefit from that additional slight top end attenuation.

It is also similar to the default Dirac Live curve which I use, again containing the more negative slope above 15k.

I could, but I don't feel like messing with target curves when I could be listening to music, as long as they have that similar smooth downward slope. I think one gets quite used to any of them within hours or days. But, while they are empirically derived from many listening tests, it is all perceptual in the end. So, go with the curve you like best.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
2,467
Likes
1,162
Location
UK
#85
I find it fascinating that the 'target curve' idea persists - and that it is probably the cause of both bad reproduction *and* bad recording. It was discussed at length in the Genelec thread but I think said discussion has had zero impact on anyone's views. If I were a 'target curve' kind of person that discussion would be a light bulb moment for me. How could I read it, understand it, and then ignore it?!

The summary is this:
  • the measured curve for any speaker and room is a consequence of the speaker's on-axis response, its dispersion characteristics, and the characteristics of the room.
  • duplicating the measured curve for another speaker/room combination makes no sense because it means an arbitrary modification of the on-axis response in order to achieve it, which will (if the direct sound is modified) sound bad even though the direct plus reflections measurement says it matches the other setup
  • a speaker should as far as possible have a flat on-axis anechoic response and uniform dispersion characteristics; this will be optimal in any room. If your speaker doesn't get near this, get another speaker.
  • minor dispersion flaws can be partially compensated for with some subtle EQ (e.g. baffle step compensation) but it's not perfect.
  • the room will do what it does; if the room is bad, change the room. You cannot just arbitrarily EQ the speaker to compensate for the room.
Floyd Toole said:
We don't feel it necessary to equalize voices and instruments for different performance spaces. Why should it be necessary to do so for loudspeakers reproducing those voices and instruments? Acoustically poor rooms can exist in both live and reproduced performances, and in neither case is equalization a solution - that is for acoustical treatments. If one starts with flawed loudspeakers, which many do, unfortunately, equalization may or may not be able to help, but the best data upon which to base such equalization is anechoic data, and if one had that, surely the best thing to do is to avoid purchasing such loudspeakers.
If I were you, I would be thinking "O. M. G. This explains everything about why I'm still trying to get a good-sounding target curve: the whole idea is bogus! Hallelujah, I can now see a way to end the pain".

But I know you won't. :)
 
Joined
May 15, 2018
Messages
24
Likes
24
Location
Neuville-sur-Saône, France
#86
How do you like to set up your house curve?
Hi,
Here is the response curve of my system, from the listening position, after equalization :

38_2018Mesure_ToScale_GDCorr.png


It is divided into three parts :
-The part above 5 kHz is not relevant. It was measured with both speakers playing at the same time, and thus the graph is very inaccurate because of possible interference between the two tweeters.
-The part between 200 and 5000 Hz follows the natural response of the speakers. Only isolated peaks or dips have been corrected, and the positive and negative corrections have been balanced.
-The part below 200 Hz is my own personal preference. I tried many things during one year, and nothing else than this could satisfy me.

The speakers are Neumann KH-120 near-field monitors. The listening distance is 2 meters, in a 3.4 x 6.2 x 2.5 meters room with no acoustic correction. They are not looking right at the listening position. They are nearly parallel. The reverberaton time at the listening position is 0.45 seconds.

The DSP correction is active from 35 to 900 Hz and performed by a MiniDSP 2x4 :


32_201706_filtres2.png



Thanks for the lengthy reply.

I would like to know your justification for saying this. I agree with it totally, but find it very hard to persuade anybody else that it is true!
Hi Cosmik. Paul Hales gives a plausible explanation in this long video


It can be summarized this way :

-The house curve is the result of the direct field and the diffuse field
-Speakers are directional : the direct field has more high frequencies than the diffuse field
-Natural sound sources, such as voice, or music instruments are directive too
-We listen to music with nearly no diffuse field recorded

=> When the speakers are perfect in anechoic conditions, because of their directivity, the house curve they give when they are not equalized mimics the sound that the real instrument would give in the same room.

...if this is right, it would mean that small speakers would be better at reproducing small instruments because of their wider directivity, and large speakers would be better for big instruments because of their narrower directivity.
 

Sal1950

Major Contributor
The Chicago Crusher
Joined
Mar 1, 2016
Messages
4,642
Likes
1,823
Location
Central Fl
#88
I think the B&K curve is probably good enough. It is close to what many more modern estimates, like Olive's and others, consider a good curve, although the modern curves tend to have an increasingly negative slope above about 15k. Not a big deal, though. But, sometimes CD playback and its potential filter artifacts benefit from that additional slight top end attenuation.
It is also similar to the default Dirac Live curve which I use, again containing the more negative slope above 15k.
I find it curious that so much attention is being given to the small differences in the slopes of various room curves above 15K? How many of todays active audiophiles can actually hear it anyway?

"The range of hearing for a healthy young person is 20 to 20,000 hertz. The hearing range of humans gets worse with age. People lose the ability to hear sounds of high frequency as they get older. The highest frequency that a normal middle-aged adult can hear is only 12-14 kilohertz. Also, the hearing range for men worsens more quickly than the hearing range for women. Christopher D'Ambrose -- 2003"
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
489
#89
Hi,
Here is the response curve of my system, from the listening position, after equalization :

View attachment 13279

It is divided into three parts :
-The part above 5 kHz is not relevant. It was measured with both speakers playing at the same time, and thus the graph is very inaccurate because of possible interference between the two tweeters.
-The part between 200 and 5000 Hz follows the natural response of the speakers. Only isolated peaks or dips have been corrected, and the positive and negative corrections have been balanced.
-The part below 200 Hz is my own personal preference. I tried many things during one year, and nothing else than this could satisfy me.

The speakers are Neumann KH-120 near-field monitors. The listening distance is 2 meters, in a 3.4 x 6.2 x 2.5 meters room with no acoustic correction. They are not looking right at the listening position. They are nearly parallel. The reverberaton time at the listening position is 0.45 seconds.

The DSP correction is active from 35 to 900 Hz and performed by a MiniDSP 2x4 :


View attachment 13282




Hi Cosmik. Paul Hales gives a plausible explanation in this long video


It can be summarized this way :

-The house curve is the result of the direct field and the diffuse field
-Speakers are directional : the direct field has more high frequencies than the diffuse field
-Natural sound sources, such as voice, or music instruments are directive too
-We listen to music with nearly no diffuse field recorded

=> When the speakers are perfect in anechoic conditions, because of their directivity, the house curve they give when they are not equalized mimics the sound that the real instrument would give in the same room.

...if this is right, it would mean that small speakers would be better at reproducing small instruments because of their wider directivity, and large speakers would be better for big instruments because of their narrower directivity.
I find it curious that so much attention is being given to the small differences in the slopes of various room curves above 15K? How many of todays active audiophiles can actually hear it anyway?

"The range of hearing for a healthy young person is 20 to 20,000 hertz. The hearing range of humans gets worse with age. People lose the ability to hear sounds of high frequency as they get older. The highest frequency that a normal middle-aged adult can hear is only 12-14 kilohertz. Also, the hearing range for men worsens more quickly than the hearing range for women. Christopher D'Ambrose -- 2003"
Well, it might not make any difference you or I could hear. But, I would like to think that these curves are derived from listening experiments on humans rather than being pulled out of thin air. Maybe they use too many younger dudes in the listening experiments. Who knows?
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
4,693
Likes
3,117
#90
I find it fascinating that the 'target curve' idea persists - and that it is probably the cause of both bad reproduction *and* bad recording. It was discussed at length in the Genelec thread but I think said discussion has had zero impact on anyone's views. If I were a 'target curve' kind of person that discussion would be a light bulb moment for me. How could I read it, understand it, and then ignore it?!

The summary is this:
  • the measured curve for any speaker and room is a consequence of the speaker's on-axis response, its dispersion characteristics, and the characteristics of the room.
  • duplicating the measured curve for another speaker/room combination makes no sense because it means an arbitrary modification of the on-axis response in order to achieve it, which will (if the direct sound is modified) sound bad even though the direct plus reflections measurement says it matches the other setup
  • a speaker should as far as possible have a flat on-axis anechoic response and uniform dispersion characteristics; this will be optimal in any room. If your speaker doesn't get near this, get another speaker.
  • minor dispersion flaws can be partially compensated for with some subtle EQ (e.g. baffle step compensation) but it's not perfect.
  • the room will do what it does; if the room is bad, change the room. You cannot just arbitrarily EQ the speaker to compensate for the room.


If I were you, I would be thinking "O. M. G. This explains everything about why I'm still trying to get a good-sounding target curve: the whole idea is bogus! Hallelujah, I can now see a way to end the pain".

But I know you won't. :)

I've made this point before also in the other thread.

If you have a perfectly flat speaker as measured anechoically, put in the room and measure it, you'll get something like the B&K curve in the room. The exact slope will vary some depending upon room size. You'll also get lots of up down wiggles from various things including reflections though above the Schroeder frequency the average through those will be about right (or with 1/6th octave smoothing).

So now we take a very non-flat speaker as measured anechoically, and put it in a room. Do we just throw up our hands and say the room is the room and changing anything EQ wise will be adverse? That would not be true. Quoting you:"You cannot just arbitrarily EQ the speaker to compensate for the room." The just arbitrarily part is what is wrong. Given what we know about windowing and measurements in room vs anechoic we know there is a perhaps poorly named "room curve" connecting those conditions. That allows us to measure a speaker that is not flat and see at least larger deviations which combined with a room curve would correct general speaker response closer to flat. It is a way to at least approach having an error curve from anechoically measuring a speaker with which compensate on playback. It isn't perfect and it can be abused, but I don't think it becomes so wrong-headed as you portray it. Especially given that many speakers are still very far from adhering closely to flat anechoic measurements.

Now optimum would be perfect speaker anechoically, measured in the spot in your room you'll put other speakers and getting a curve for in room measures. Even then it might vary slightly due to directional differences between 'perfect' speakers and other speakers. I think these sorts of differences in directionality between designs, and room positions and room size is why the exact B&K or other curve is not correct in any given situation. So altering the exact curve can sometimes at least please the listener more in terms of overall balance in frequency response. I do agree with you that me creating the curve that pleased me in my room really tells you very little about any specifics for your room or speakers. Having done or helped with room EQ for long time now this matches my experiences. I find you need to look at two areas. The steepness of the slope and where to let it shelve out in the low end. The moving of the shelving area lower in larger rooms seems to work. The steepening of the slope in smaller rooms seems to hold most of the time.

So I do have a question for you, if you had a speaker very unflat anechoically are you saying fixing that response with EQ wouldn't improve the result in your room?

Also, perhaps Dr. @Floyd Toole could tell me if I've got this wrong. I wish all speakers were designed flat, but that isn't the case yet.
 

Sal1950

Major Contributor
The Chicago Crusher
Joined
Mar 1, 2016
Messages
4,642
Likes
1,823
Location
Central Fl
#92
Well, it might not make any difference you or I could hear. But, I would like to think that these curves are derived from listening experiments on humans rather than being pulled out of thin air. Maybe they use too many younger dudes in the listening experiments. Who knows?
So I guess it depends on what illusion we want to recreate. The one we heard at the concert last week, or the one we heard in our 20-30s. Maybe if we want to feel younger the house curve should be flatter? :)
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
2,467
Likes
1,162
Location
UK
#93
So I do have a question for you, if you had a speaker very unflat anechoically are you saying fixing that response with EQ wouldn't improve the result in your room?
No, I am not saying that. But I am saying that the in-room response is the wrong curve to be looking at. It is the anechoic measurement that would be key here: set that up to be flat. But...

...The other aspect is dispersion. If this is a really terrible speaker in this regard, then something has to be done about that in the form of partial compensation that will necessarily damage the on-axis (direct) response - which will be heard separately regardless of the way it seems to 'correct' the in-room curve. Baffle step compensation is a compromise that works for a decent speaker with smoothly changing dispersion. This can be applied by a combination of calculation (the curve shape) based on baffle size and shape, and by ear (the depth) without reference to frequency response measurements.

The in-room curve is the single dimensional reduction of several dimensions. Your hearing discerns the individual dimensions; your eyes looking at the simplified curve cannot!

The easiest thing, of course, is to avoid speakers that have poor dispersion or other ills that cannot be corrected, but the in-room response cannot tell you how your speaker is performing in that regard.

I think we are basically agreeing, but you take a more charitable, less absolutist approach than I do :).
 
Last edited:

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
489
#94
So I guess it depends on what illusion we want to recreate. The one we heard at the concert last week, or the one we heard in our 20-30s. Maybe if we want to feel younger the house curve should be flatter? :)
Well, if you could only remember the fine details of what stuff sounded like when younger, that might be something of a strategy. Except, we are kidding ourselves if we think we do remember. So, we have to do the best we can with the ears we have now. And, if we can't hear too much into that top octave due to age, it doesn't matter much anyway. In any case, I think going with our best recollection of a recent live event is going to be much better, albeit imperfect due to acoustic memory and all that.

I don't think the slight rolloff in the top octave is too much of a big deal, as I, an old man, said. But, it is worth experimenting with subjectively over a wide range of recordings. The Pro version of your Audyssey, which I used to use, had a choice from a family of three optional room curves, each with progressively more top octave rolloff starting at a lower frequency. They were there to allow choosing the curve that best fit your room volume, and that was quite audible to me. Acoustics professionals have totally accepted this idea for a long time and use it routinely in EQing different sized halls, cinemas or HT setups.

The curve for the smallest room was technically correct and definitely preferable to me sonically in Audyssey. And, incidentally, a totally flat curve with no downward slope or hi end rolloff was definitely not preferable long before I read Olive or anyone else on the subject. Also, Audyssey's "midrange compensation" dip was definitely not a winner in my system or others I heard with it. I think Audyssey missed the boat on that one. Sean Olive and others agree. But, you can now edit to flatten that out, as I once used to do in Audyssey Pro.

So, established target curves based on scientific research are a good place to start. B&K's pioneering efforts are somewhat old, although the main audible difference from today's curves due to the top octave is slight, I believe. They did, however, nail the general downward sloping idea, though some still may quibble over tenths of dB per octave in the slope. I also believe newer experimentally-derived curves using newer equipment and test setups are likely generally better as perceived by the general population.

But, if you have the patience, modify the target curve to your heart's content subjectively. As I said, though, I grew tired of doing that, and I found that the unmodified Dirac curve was quite satisfying over a wide range of classical recordings using your recent live concert method.

Dirac, unlike Audyssey or many other tools, is also nice in that you can set up 4 different curves and instantaneously switch between them. A friend does that with different program material - one for CD, one for SACD, etc. Others use that for changes to the room environment, etc. - drapes open/closed, screen up/down, boosted bass shelving on/off, etc. I don't have the need or desire to do that, myself.
 

Sal1950

Major Contributor
The Chicago Crusher
Joined
Mar 1, 2016
Messages
4,642
Likes
1,823
Location
Central Fl
#95
Dirac, unlike Audyssey or many other tools, is also nice in that you can set up 4 different curves and instantaneously switch between them.
That's a nice feature. The new Audyssey Editor app allows me to create as many different curve files as I want but it takes a couple minutes to upload the large files to the AVR to make the change. I take it that after install of Pro there must have been some means for the AVR to store multiple files internally, interesting.
Editor offers the same options for the top end rolloff selections and makes the midrange dip switchable. I can also customize the full range curves for each individual speaker any way I chose. Nice little app that replaced Pro for only $20 and you don't have to hire a professional with the license to use it. ;)
 

Floyd Toole

Industry Insider
Technical Expert
Industry Insider
Joined
Mar 12, 2018
Messages
85
Likes
295
#96
I've made this point before also in the other thread.

If you have a perfectly flat speaker as measured anechoically, put in the room and measure it, you'll get something like the B&K curve in the room. The exact slope will vary some depending upon room size. You'll also get lots of up down wiggles from various things including reflections though above the Schroeder frequency the average through those will be about right (or with 1/6th octave smoothing).

So now we take a very non-flat speaker as measured anechoically, and put it in a room. Do we just throw up our hands and say the room is the room and changing anything EQ wise will be adverse? That would not be true. Quoting you:"You cannot just arbitrarily EQ the speaker to compensate for the room." The just arbitrarily part is what is wrong. Given what we know about windowing and measurements in room vs anechoic we know there is a perhaps poorly named "room curve" connecting those conditions. That allows us to measure a speaker that is not flat and see at least larger deviations which combined with a room curve would correct general speaker response closer to flat. It is a way to at least approach having an error curve from anechoically measuring a speaker with which compensate on playback. It isn't perfect and it can be abused, but I don't think it becomes so wrong-headed as you portray it. Especially given that many speakers are still very far from adhering closely to flat anechoic measurements.

Now optimum would be perfect speaker anechoically, measured in the spot in your room you'll put other speakers and getting a curve for in room measures. Even then it might vary slightly due to directional differences between 'perfect' speakers and other speakers. I think these sorts of differences in directionality between designs, and room positions and room size is why the exact B&K or other curve is not correct in any given situation. So altering the exact curve can sometimes at least please the listener more in terms of overall balance in frequency response. I do agree with you that me creating the curve that pleased me in my room really tells you very little about any specifics for your room or speakers. Having done or helped with room EQ for long time now this matches my experiences. I find you need to look at two areas. The steepness of the slope and where to let it shelve out in the low end. The moving of the shelving area lower in larger rooms seems to work. The steepening of the slope in smaller rooms seems to hold most of the time.

So I do have a question for you, if you had a speaker very unflat anechoically are you saying fixing that response with EQ wouldn't improve the result in your room?

Also, perhaps Dr. @Floyd Toole could tell me if I've got this wrong. I wish all speakers were designed flat, but that isn't the case yet.
The complete answer is in many pages in my book, but to reduce the argument to its elements there are four facts to remember.
First, an omni mic and analyzer are not equivalent to two ears and a brain. Rooms curves show evidence of phenomena that are not problems for binaural hearing humans. Equalizing these phenomena can degrade good loudspeakers.
Second, from comprehensive anechoic data on a loudspeaker (the spinorama for example) one can predict with good accuracy the steady state room curve in a typically reflective room. However the reverse is not true. The spinorama can identify good or bad loudspeakers - this capability is seriously compromised if in-room measurements are all that is possible.
Third, a loudspeaker that is not flat on axis but which has well-behaved directivity as a function of frequency can benefit from in-room equalization, but in order to know that one needs comprehensive anechoic data. If one had such data, the optimum equalization (above the transition frequency) would be based on the anechoic data, not a room curve. Also, if one had the anechoic data, one should not have purchased the loudspeaker to begin with.
Fourth, about 30% of our overall opinion of sound quality is attributable to bass performance - not only of the loudspeaker, but of the individual room acoustics through which it is communicated to listeners. EQ is one tool to improve the situation, especially if one employs multiple subs in a sensible configuration, as discussed in tiresome detail in Chapter 8. Steady-state room curves are the metric for this part of the frequency range and it should be smooth. Tilt, if any, is a matter of taste for the program being played. The "circle of confusion" is especially active in this part of the frequency range so a bass tone control is useful, especially if one enjoys older recordings.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
4,693
Likes
3,117
#97
The complete answer is in many pages in my book, but to reduce the argument to its elements there are four facts to remember.
First, an omni mic and analyzer are not equivalent to two ears and a brain. Rooms curves show evidence of phenomena that are not problems for binaural hearing humans. Equalizing these phenomena can degrade good loudspeakers.
Second, from comprehensive anechoic data on a loudspeaker (the spinorama for example) one can predict with good accuracy the steady state room curve in a typically reflective room. However the reverse is not true. The spinorama can identify good or bad loudspeakers - this capability is seriously compromised if in-room measurements are all that is possible.
Third, a loudspeaker that is not flat on axis but which has well-behaved directivity as a function of frequency can benefit from in-room equalization, but in order to know that one needs comprehensive anechoic data. If one had such data, the optimum equalization (above the transition frequency) would be based on the anechoic data, not a room curve. Also, if one had the anechoic data, one should not have purchased the loudspeaker to begin with.
Fourth, about 30% of our overall opinion of sound quality is attributable to bass performance - not only of the loudspeaker, but of the individual room acoustics through which it is communicated to listeners. EQ is one tool to improve the situation, especially if one employs multiple subs in a sensible configuration, as discussed in tiresome detail in Chapter 8. Steady-state room curves are the metric for this part of the frequency range and it should be smooth. Tilt, if any, is a matter of taste for the program being played. The "circle of confusion" is especially active in this part of the frequency range so a bass tone control is useful, especially if one enjoys older recordings.
Thank you for the information answering my question.
 

Similar threads

Top Bottom