What is your favorite house curve

[mitchco]

There is more to it than just a "house curve". Here are a few industry guidelines:

In 1974, Bruel and Kjaer instruments wrote an application note on, "Relevant loudspeaker tests in studios in
Hi-Fi dealers' demo rooms in the home etc. using 1/3 octave, pink-weighted, random noise." See Figure 5:
"The optimum curve for HIFI equipment measured in the actual listening room." While the curve is smooth, it
does have a gradual roll-off. Looking at the frequency response plot, the gradual roll-off starts at -0.5 dB at
200 Hz, -3 dB at 2 kHz, and -6 dB down at 20 kHz. Over time, this became known as the B&K "house
curve." For decades, professionals and amateurs alike have used the B&K house curve (or target response)
when equalizing a sound reproduction system for critical listening purposes.

In 1994 (updated in 2015), the ITU produced a "Recommendation ITU-R BS.1116-3 (02/2015) Methods for the
subjective assessment of small impairments in audio systems." On Page 17, section 8.3.4.1 Operational
room response curve shows a tolerance limit of ±3 db SPL and a very similar operational room response
curve as per EBU-Tech 3276 below.

In 1998, the European Broadcast Union produced a Tech note, EBU-Tech 3276, “Listening conditions for the
assessment of sound programme material: monophonic and two–channel stereophonic." See Figure 2 on
Page 6: "Tolerance limits of the operational room response curve." The curve is flat to 2 kHz and then a
straight line to -6 dB at 20 kHz with a ±3 dB tolerance along the target. As mentioned previously, this Tech
note has other target specifications that we will examine in upcoming sections.

In 2009, Dr. Sean Olive from Harmon wrote a paper on, "The Subjective and Objective Evaluation of Room
Correction Products." Presentation download. Through the use of measurements and listening tests, the
preferred response is a straight line, but tilted starting at +2 dB at 20 Hz to -8 dB at 20 kHz. The conclusion
is that flat in-room response is not the preferred target response.

In 2015, Dr. Floyd Toole wrote an AES open access paper on, "The Measurement and Calibration of Sound
Reproducing Systems." See Figure 14 "Subjectively preferred steady-state room curve targets in a typical
domestic listening room [49], from Olive et al. [48]. The prediction of Fig. 13(a) is superimposed." Note that
the trained listener's preference target curve is similar to the other targets referenced above.

Overlaying the "house curves" or target responses:

Coincidence... Personally, I favour the Olive/Toole house curve (or really, tilt). I have tried most of those "curves" in many, many rooms, including several recording studio control rooms, with many speakers, including live sound reinforcement systems over a... gulp... 38 year time span. Starting with a 10 band analog eq all the way to state of the art DSP, as can be seen in my sig.

While there are other contributing factors, i.e. physical shape of the room, how live or dead, one can't go too wrong by starting with any one of these house curves. Especially, if one is using modern DSP, each one can be compared in real time and fined tuned to one's personal preference.

 

[DonH56]

That is awesome Mitch, thanks!

My room is very dead and my curve is close to the old B&K curve. May be subconscious since that's the one I was weaned on. I find the newer ones that are flatter a little higher in the midrange tend to put a bit of upper-midrange "bite" that is exacerbated by many pop recordings that seem a bit too bright already. Probably just me.

 

[amirm]

Here is the curve applied by different EQs references in mitcho's post:

Dashed line is no EQ. The prefered ones were the ones sloping down (red and green).

 

[Cosmik]

Personally, I favour the Olive/Toole house curve (or really, tilt).

How do you know?

If your speakers are anything like a normal speaker they feature a 'baffle step' where dispersion transitions from directional to omni as you go downwards, and they might - loosely - feature narrowing dispersion at higher frequencies, but it varies for two and three way versions, panels, BMRs, horns, etc.

These are deviations from neutral and cannot be 'corrected' per se if you are listening in the far field because the direct and ambient sounds are different. However, they can be partially compensated for subjectively by 'compromise'. I certainly do it for my speakers (which are monkey coffins and as such will behave differently from modern narrow-fronted boxes). At least in the DSP case, we get to maintain linear phase for the direct sound.

I think this downward tilt thing is this compensation being mistaken for a target when, in fact, it is just a consequence of different speakers' dispersion characteristics combined with the room, and the subjective compensation thereof. And of course it means that there is no 'ideal' tilt because it depends on your speakers' characteristics and their position in the room, etc. - or should do. It is clear that there are people taking home their new Kii Threes and 8c and excitedly dialling in their 'house curve' only to find they don't sound subjectively right (but they'll persevere anyway) even though they seem to measure the same as other high end speakers in the same room. In fact, they probably would have sounded much better just 'out of the box'.

 

[NorthSky]

Mitchco is simply mentioning, from a 38-year time span of experimenting.

 

[mitchco]

That is awesome Mitch, thanks!

My room is very dead and my curve is close to the old B&K curve. May be subconscious since that's the one I was weaned on. I find the newer ones that are flatter a little higher in the midrange tend to put a bit of upper-midrange "bite" that is exacerbated by many pop recordings that seem a bit too bright already. Probably just me.

Hey Don! Yes, I grew up on the B&K curve as well. I was working in a number of recording studios at the time where the large format monitors were eq’d to the B&K curve. I really like the tonal balance. Sounds neutral to my ears, without the hyped top end that drives my ears crazy. Nope it ain't just you man

Generally speaking to the forum, if one digs into the industry guidelines linked above (there are a bunch more if you search for it), there are many other recommended specifications from early reflections, room decay, speaker directivity, setup height and angles, monitoring level etc. It is not just one item, but a collection of specifications that yields a system capable of meeting the measured specifications of a critical listening environment as defined in these guidelines. Blueprints if you will, and they are all more similar than dissimilar.

Coming from working in many different recording studio control rooms, one becomes very grateful when the room is built and calibrated to a standard. Why? Imagine you got to record, mix and master your favorite band. It was awesome, great recording. But now you move to a different facility to mix down this great recording. When you left the facility, you had laid down a “scratch mix” which is a hastily put together mix for each multi-track song. Levels, eq, effects, are all documented and some are automated. So you know, sounds half decent, the band members make cassettes to listen… everyone pumped.

You arrive at the new studio with the band, everyone is excited to mix this great recording. Also, the money guys and studio execs are there to greet the band and listen to one of the scratch mixes, while you set up. Tape goes on the deck, settings get hastily put into each channel strip on the mixing console for at least 24 tracks, if not 48, eq, levels, effects…

Roll tape! And … it sounds nothing like your scratch fix you had from the previous studio. The eq on every track sounds wrong, each fader level sounds wrong, and the tonal balance is way off. It sounds like shite! The band is looking nervous, the money guys are looking at the record execs, who are looking at you… in an uncalibrated control room….

Or Roll tape! Oh yeah!!! Sounds fantastic in the first 5 seconds, you crank the monitors and the drinks come out and the party begins. Great performance sounds fantastic!

I lived both of those stories more than once.

But there is the rub.

Each measurement software I have used, like REW, ARTA HolmImpulse, rePhase, Dirac, Acourate, and Audiolense, have their own windowing and smoothing algorithms, plus some have frequency dependent windowing and proprietary psychoacoustic filtering. They are all close, but each on is distinctively different. Because of this, the designed “house curve” in each software needs to be slightly different to achieve the same result using a baseline program like REW to verify.

For example, if I want to hear the same tonal response using Audiolense as I do using Acourate, the two targets need to be slightly different, in order to produce the same result when both measured in REW.

Additionally, the B&K curve is based on 1/3 octave pink noise, which is much different than the DSP stimulus and analysis software used today. However, one can figure out what the target has to look like in the design software in order to produce the B&K curve, and validating using the same measurement technique as described in the application note.

In the end, we are all probably listening to something that is more similar than dissimilar, regardless of what the house curve looks like in the design software, unless it is a huge deviation…

 

[mitchco]

How do you know?

If your speakers are anything like a normal speaker they feature a 'baffle step' where dispersion transitions from directional to omni as you go downwards, and they might - loosely - feature narrowing dispersion at higher frequencies, but it varies for two and three way versions, panels, BMRs, horns, etc.

These are deviations from neutral and cannot be 'corrected' per se if you are listening in the far field because the direct and ambient sounds are different. However, they can be partially compensated for subjectively by 'compromise'. I certainly do it for my speakers (which are monkey coffins and as such will behave differently from modern narrow-fronted boxes). At least in the DSP case, we get to maintain linear phase for the direct sound.

I think this downward tilt thing is this compensation being mistaken for a target when, in fact, it is just a consequence of different speakers' dispersion characteristics combined with the room, and the subjective compensation thereof. And of course it means that there is no 'ideal' tilt because it depends on your speakers' characteristics and their position in the room, etc. - or should do. It is clear that there are people taking home their new Kii Threes and 8c and excitedly dialling in their 'house curve' only to find they don't sound subjectively right (but they'll persevere anyway) even though they seem to measure the same as other high end speakers in the same room. In fact, they probably would have sounded much better just 'out of the box'.

Hello Cosmik,

I know because I have had a 2-way constant high directivity loudspeaker, 3 way constant directivity, a cone and dome, a wide dispersion and a panel speaker (Maggie 2.6R but data lost on hard drive). I used the same house curve on all of them with excellent results. So while I hear you on that, my ears and measurements say it is not the case. Sure there is some difference, but does not account for the downward tilt on all of them... In fact, what I am finding is that most of these out of the box speakers are way too bright to begin with. Let's see if that is the case with the 8c's on their way for review...

While I am a total directivity fan, I think there is actually more variation between the DSP design software than there is directivity in well-designed speakers, as mentioned in my reply to Don’s post. Man, those JBL 4722’s I have are really high directivity:

Yet, I used the same house curve on the Phantom Golds and happy with the tonal balance of both. If I overlay my REW measures with Michael's M2's we are both listening to a very similar tonal response. Same with members of other forums and you will see a link in one of the articles above to a 25 speaker line array, yet virtually the same tilted REW measured response at the listening position, from half way around the world.

It is fairly easy to dial in a house curve that is too bright and one that is too dull. One can then "bracket" house curves in 1 dB increments of adjusting the tonal response and generate a half dozen correction filters between too bright and too dull. One can then switch between each one in real time in JRiver for example and very quickly narrow down ones preference or speaker voicing, that fits well within industry guidelines, but is your personal preference.

It took me a few hours of listening to get it down to 1 or 2 candidates and then over a period of days, I listened to a wide variety of music and video material and settled on a house curve that worked well on virtually everything. Sounds neutral to my ears and not overly bright on anything - my preference. While accounting for differences in the DSP design software, it is still the same house curve, independent of speaker directivity.

 

[Wombat]

Hello Cosmik,

I know because I have had a 2-way constant high directivity loudspeaker, 3 way constant directivity, a cone and dome, a wide dispersion and a panel speaker (Maggie 2.6R but data lost on hard drive). I used the same house curve on all of them with excellent results. So while I hear you on that, my ears and measurements say it is not the case. Sure there is some difference, but does not account for the downward tilt on all of them... In fact, what I am finding is that most of these out of the box speakers are way too bright to begin with. Let's see if that is the case with the 8c's on their way for review...

While I am a total directivity fan, I think there is actually more variation between the DSP design software than there is directivity in well-designed speakers, as mentioned in my reply to Don’s post. Man, those JBL 4722’s I have are really high directivity:

View attachment 10883

Yet, I used the same house curve on the Phantom Golds and happy with the tonal balance of both. If I overlay my REW measures with Michael's M2's we are both listening to a very similar tonal response. Same with members of other forums and you will see a link in one of the articles above to a 25 speaker line array, yet virtually the same tilted REW measured response at the listening position, from half way around the world.

It is fairly easy to dial in a house curve that is too bright and one that is too dull. One can then "bracket" house curves in 1 dB increments of adjusting the tonal response and generate a half dozen correction filters between too bright and too dull. One can then switch between each one in real time in JRiver for example and very quickly narrow down ones preference or speaker voicing, that fits well within industry guidelines, but is your personal preference.

It took me a few hours of listening to get it down to 1 or 2 candidates and then over a period of days, I listened to a wide variety of music and video material and settled on a house curve that worked well on virtually everything. Sounds neutral to my ears and not overly bright on anything - my preference. While accounting for differences in the DSP design software, it is still the same house curve, independent of speaker directivity.

The above 10k drop-off is interesting wrt those who believe they can hear above above 20k affects.

 

[mitchco]

Note how far away the measurement was taken and that it is inside an arena... but I am not going to argue your point

 

[Wombat]

Note how far away the measurement was taken and that it is inside an arena... but I am not going to argue your point

Missed that. Interestingly. or otherwise, I tend to listen at 5 to 6 metres to my Altec system but by no means in a hall, let alone an arena.

 

[RayDunzl]

Listening to The Northern Pikes - Snow in June

"additional recording: Mitch Barnett"

I wonder if you're lurking in anything else in the rack...

 

[dallasjustice]

There’s no perfect target. I’m sure there’s some variations based on room, speaker, genre and personal preference. However, I’ve noticed that a downward tilt is best for me. You can see it here:

The best way is to experiment on your own. With Roon it’s easy to load different targets into different playback zones and instantly switch between them with Roon remote; less than a second delay between switch. Here’s how:

I set one zone as ASIO output and the other WASAPI. (WASAPI and ASIO are identical) Each output gets their own filter in DSP.

Here I’ve loaded the identical targets in each zone. The only difference is that one target is a minimum phase and the other is linear phase.

 

[Sal1950]

My favorite house curve,

 

[Jerry Sobel]

Ok, here's my house curve at the moment. Listening to HDRadio WSMR Classical on the LSR 308.

View attachment 10872

Interesting that you dropped the bass frequencies. There is som every good reading on AV Forums on EQ curves.

 

[Fitzcaraldo215]

Hello Cosmik,

I know because I have had a 2-way constant high directivity loudspeaker, 3 way constant directivity, a cone and dome, a wide dispersion and a panel speaker (Maggie 2.6R but data lost on hard drive). I used the same house curve on all of them with excellent results. So while I hear you on that, my ears and measurements say it is not the case. Sure there is some difference, but does not account for the downward tilt on all of them... In fact, what I am finding is that most of these out of the box speakers are way too bright to begin with. Let's see if that is the case with the 8c's on their way for review...

While I am a total directivity fan, I think there is actually more variation between the DSP design software than there is directivity in well-designed speakers, as mentioned in my reply to Don’s post. Man, those JBL 4722’s I have are really high directivity:

View attachment 10883

Yet, I used the same house curve on the Phantom Golds and happy with the tonal balance of both. If I overlay my REW measures with Michael's M2's we are both listening to a very similar tonal response. Same with members of other forums and you will see a link in one of the articles above to a 25 speaker line array, yet virtually the same tilted REW measured response at the listening position, from half way around the world.

It is fairly easy to dial in a house curve that is too bright and one that is too dull. One can then "bracket" house curves in 1 dB increments of adjusting the tonal response and generate a half dozen correction filters between too bright and too dull. One can then switch between each one in real time in JRiver for example and very quickly narrow down ones preference or speaker voicing, that fits well within industry guidelines, but is your personal preference.

It took me a few hours of listening to get it down to 1 or 2 candidates and then over a period of days, I listened to a wide variety of music and video material and settled on a house curve that worked well on virtually everything. Sounds neutral to my ears and not overly bright on anything - my preference. While accounting for differences in the DSP design software, it is still the same house curve, independent of speaker directivity.

Good stuff, Mitch. And, I totally agree about the downward sloping curve.

But, I thought at least part of the problem was a measurement issue, owing to the fact that omni measurement mikes pick up HF reflected energy that the ear does not to the same degree, due to its shape, contours, placement on the side of the head, pinnae, etc.

 

[Jerry Sobel]

When I do perform localized bass optimization with PEQ filters, after I am done I put in a shelf boost of the whole region to taste. Without it, there is a loss of bass impact.

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.

Curious that it drops the bass frequencies. From what I have read humans prefer a relative bass boost when compared to the higher frequencies. How does it sound? That is the bottom line.

 

[Jerry Sobel]

There’s no perfect target. I’m sure there’s some variations based on room, speaker, genre and personal preference. However, I’ve noticed that a downward tilt is best for me. You can see it here:
View attachment 10896 View attachment 10895
The best way is to experiment on your own. With Roon it’s easy to load different targets into different playback zones and instantly switch between them with Roon remote; less than a second delay between switch. Here’s how:
View attachment 10897 View attachment 10898
I set one zone as ASIO output and the other WASAPI. (WASAPI and ASIO are identical) Each output gets their own filter in DSP.
View attachment 10899 View attachment 10900
Here I’ve loaded the identical targets in each zone. The only difference is that one target is a minimum phase and the other is linear phase.

Thanks!

 

[DonH56]

The baseline curves do not reduce the bass; look again at the curves in Mitch's post. The in-room responses Amir showed do so that may be the confusion; everything rolls off somewhere, and most RC programs will not boost bass below what it decides is the LF cut-off of the speakers in the room as measured.

The bass in my room rolls off too, below about 7 Hz.

 

[Jerry Sobel]

The baseline curves do not reduce the bass; look again at the curves in Mitch's post. The in-room responses Amir showed do so that may be the confusion; everything rolls off somewhere, and most RC programs will not boost bass below what it decides is the LF cut-off of the speakers in the room as measured.

The bass in my room rolls off too, below about 7 Hz.

Right you are Senor

 

[RayDunzl]

Curious that it drops the bass frequencies. From what I have read humans prefer a relative bass boost when compared to the higher frequencies.

As I read it:

Equal loudness curves, at least the early ones, converge in the bass region - need higher SPL levels at lower frequencies at lower midrange SPL for equal loudness, until you get to 90 or 100dB, then the curve is pretty flat.

That, 90-15dB, is my (peak) attentive listening area, so I tried no boost for bass. Let the recording come through. And maybe listen to the results of early investigators that hadn't been subjected to modern "bass". I don't dislike how it sounds turned down, so, ???

Newer Equal Loudness Curves (ISO) show bass boost even at the higher SPL levels.

Fletcher Munson - 1933 (maybe)

ISO - 2003 (maybe)

25db boost at 31.5Hz with 90dB midrange?

No thanks. I tried it. Mud here.

YMMV.