• 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!

What is your favorite house curve

Hello friends,

As a followup of my above post #300,,,

Now, just for your reference, let me share my typical "house curve" measured by my rather naive and primitive "cumulative white noised averaging method" (refer to my post here for the details of my multichannel multi-amplifier setup).
WS003807.JPG


As for the upward Fq response above 6 kHz (flexibly adjustable on-the-fly) compensating my age dependent slight hearing decline, please refer to here and here.

Also please note that I have established 0.1 msec precision time alignments between all the SP drivers as well as L-SPs and R-SPs (please refer to my summary post here) which cannot be shown in above simple Fq response house curve.
 
Last edited:
Does this mean the "Midrange Compensation" dip that Audyssey recommends around 2kHz is not a good idea for well measuring speakers because it is changing the speaker's characteristics above 500Hz? How about the "High Frequency Roll-off" that they recommend?
 
Does this mean the "Midrange Compensation" dip that Audyssey recommends around 2kHz is not a good idea for well measuring speakers because it is changing the speaker's characteristics above 500Hz? How about the "High Frequency Roll-off" that they recommend?
Do you have a source for these recommendations?
 
Does this mean the "Midrange Compensation" dip that Audyssey recommends around 2kHz is not a good idea for well measuring speakers because it is changing the speaker's characteristics above 500Hz? How about the "High Frequency Roll-off" that they recommend?

Whether or not a visible 2kHz xo dip needs to be compensated depends in the speaker and your subjective listening assessment — same goes for any other broad EQ like HF shelving filters. Sometimes it’s best left alone since the speaker designer likely already accounted for it. A quick A/B listening test with different PEQs is really not that hard to do — it may be faster and easier to accomplish when done on a PC directly than through an AVR, though. Instead of continually guessing, go apply the filters in real time and listen with your own ears!
 
Does this mean the "Midrange Compensation" dip that Audyssey recommends around 2kHz is not a good idea for well measuring speakers because it is changing the speaker's characteristics above 500Hz? How about the "High Frequency Roll-off" that they recommend?
Yes it does - it is a terrible idea, as it is based a specific flaw, common on a specific type (quite common) speaker...

If your speakers does not cross-over at that frequency, or does not have that crossover flaw - then it just degrades the sound.

First thing most people with Audyssey should do is turn MRC off. (sadly my previous AVR with Audyssey, had no option to do so!)
 
Yes it does - it is a terrible idea, as it is based a specific flaw, common on a specific type (quite common) speaker...

If your speakers does not cross-over at that frequency, or does not have that crossover flaw - then it just degrades the sound.

First thing most people with Audyssey should do is turn MRC off. (sadly my previous AVR with Audyssey, had no option to do so!)
Yes. I don’t even know why Audyssey still keeps that nonsense on by default after years of feedback to switch it off.
 
I also like these house curves.:)
WS00005257.JPG
 
Yes. I don’t even know why Audyssey still keeps that nonsense on by default after years of feedback to switch it off.
Perhaps because they sold it to Sound United years ago - it's not theirs any more.

Sound United (Masimo / Marantz-Denon) - hasn't got the people that developed Audyssey - they stayed with the company

Hence, no development for 10 years - even the MultEQ XT app is just a new user interface on the old Audyssey Pro software.

My old 2013 AVR with Audyssey XT32, has exactly the same Audyssey as the latest D&M AV1.... (but I would need to find a vintage Audyssey Pro Kit to disable MRC.... they're a bit hard to find now!)
 
This is excellent advice, and if one wishes, it can be elaborated to estimate off-axis performance as well, simply by rotating the loudspeaker. In the extreme, it is possible to generate the data necessary to create a spinorama, but only for higher frequencies. Because of the time windowing the frequency resolution becomes so large as to be useless at lower frequencies. In the spinorama database there are examples of people who appear to have done this. One also sees examples of excessive smoothing - low resolution in the frequency domain - probably done to hide evidence of resonances that would be embarrassing. So, just seeing a spinorama does not mean that it is trustworthy evidence. People lie . . . surprise, surprise.

Dr. Toole,

Thank you for chiming in on ASR. That's a real treat, you are one reason for engineers, or anyone to feel good about being Canadian!

I have a question not related to the topic on hand but hope you don't mine commenting regardless. You have said before that comparison listening tests should be done blind and if one knows which one is playing, you don't care what one thinks about which one sound better..., just going by memory that may not be what you meant. If you did mean it that way, would those comments apply to comparing many of the well measured audio amplifiers including AV receivers and AV processors? I don't recall you ever say much about audio electronics in you book, just wonder why?
 
OK. I thought about opening a rat's nest by doing this, because it is a simplification of what sometimes has to be done. Some of the points have been made earlier in this forum thread and elsewhere, but it might be useful to bring the key factors in the process to one place. The marketing of room EQ algorithms often presents the impression that all combinations of loudspeakers and rooms can be "fixed", "calibrated" or the like, by means of measurements, math and equalization. In reality, much of the "math" does not include the exceptionally complex, non-linear and occasionally capricious psychoacoustics of human listeners. A critical missing element is that humans adapt to circumstances, bringing our perceptions into acceptable territory. Loudspeakers reproduce sounds. Musicians produce sounds. Both do it in rooms. We don't feel the need to "equalize" - even if we could - the instruments and voices of live music. Two ears and a brain separate the sources from the venue, and adapt to aspects of what the environment contributes to the overall performance. The venues vary, and some are even not ideal, but we manage to appreciate the excellence of fine instruments and voices in most of them.

The special problem with sound reproducing systems is that flaws get superimposed on everything that is played through them. These monotonous colorations can sometimes be beyond the ability of humans to adapt, and they need to be identified and attenuated.

Therefore, the "right way" begins with choosing well designed, timbrally neutral, loudspeakers. If the loudspeakers exhibit audible resonances and/or frequency-dependent directivity issues, it is not likely that measurements in a room will reveal such problems and that equalization is capable of compensating for them. It is often the case that the solution is better loudspeakers. Fortunately these can be identified with good reliability from competently made anechoic measurements presented in a "spinorama" format, following the industry standard. Amir, on this site, makes such measurements and others can be found at www.spinorama.org.

This done, set them up in your room and make a steady-state frequency response measurement at the prime listening position - the stereo seat. We will be paying close attention to the frequencies below about 400-500 Hz, where adjacent boundary effects and room resonances are active. Because much of the bass in recordings is mono (all of it in LPs) drive both loudspeakers simultaneously to evaluate what is happening at low frequencies. Measure them individually to find out what is happening at frequencies above about 400 Hz. If you are using bass management and one or more subwoofers the process is the same, and of course all subs should be running simultaneously. Why? Because multiple sound sources couple energy to room resonances differently when they operate in unison.

You can repeat this at different seats to see how much seat-to-seat variation there is - often quite a lot. Averaging several of these curves is a common practice, making the curves look much smoother, but hiding some awkward realities at low frequencies. Superimposing the curves on one graph is a more useful display of what is happening in your setup. You can then choose which humps/peaks to attenuate, depending on which seats are affected. Remember, at this stage we are looking only at bass frequencies. Narrow dips, however deep, should be ignored. Broad dips can be filled in, but keep the EQ boosts below about 6 dB. Aim for a smoothish curve that is tilted slightly upward at lower frequencies.

The benefits of this exercise will apply only to the seat or seats exhibiting similar shaped curves. That is why multiple-sub methods have been developed aimed at reducing seat-to-seat variations so that one equalization can deliver improved bass to several listeners. These are discussed in detail in Chapter 8 in the 3rd edition of my book.

Above about 400-500 Hz the "early reflections" curve in the spinorama should be similar to what you have measured. If you have well designed loudspeakers the room curve might have some smallish ripples caused by acoustical interference between and among the direct and reflected sounds - these are not problems to two ears and a brain and equalization is the wrong method of addressing them if they were - that is an acoustics issue. Spatial averaging over several microphone locations tends to smooth the room curve at middle and high frequencies, thereby reducing the likelihood that an auto-EQ algorithm (or a person) might try to "fix" something that can't be fixed, or that doesn't need to be fixed. Remember, any EQ applied to a room curve modifies the direct sound, and it the the direct sound that is a key factor in determining sound quality. If you began with loudspeakers designed to have the desirable smooth and flat on-axis/listening window response, they will be degraded.

Finally, pay attention to the overall shape of the room curve. Usually, at least for conventional forward-firing loudspeakers, the room curve will tilt gently downward. If the shape deviates substantially from the early-reflections spinorama curve then one can suspect something is amiss in the acoustical treatment of the room. If listening confirms a problem, then one is free to try modifying the shape of the spectrum with broadband, low-Q, tone-control kinds of equalization. When listening to recordings we get into the circle-of confusion dilemma, where it is difficult to know where the problem lies: the playback system or the recording.

Don't worry about little ripples. When I see exceptionally smooth high-resolution room curves I strongly suspect that something wrong has been done. The measurement microphone is no substitute for two ears and a human brain. Happy landings!
That post should be a pinned thread on it's own and mandatory for reading right after signing in ASR.
Even better given as a brochure with every speaker someones buys.
Thank you Dr Toole!
 
Perhaps because they sold it to Sound United years ago - it's not theirs any more.

Sound United (Masimo / Marantz-Denon) - hasn't got the people that developed Audyssey - they stayed with the company

Hence, no development for 10 years - even the MultEQ XT app is just a new user interface on the old Audyssey Pro software.

My old 2013 AVR with Audyssey XT32, has exactly the same Audyssey as the latest D&M AV1.... (but I would need to find a vintage Audyssey Pro Kit to disable MRC.... they're a bit hard to find now!)
As far as I know. See their website. Audyssey exists as sub entity of SoundUnited with their own staff / development team, which frequently releases updates.

So yes there would have been opportunity to change the default setting for that wretched “BBC dip”.

Or does SoundUnited sell speakers with that flaw? That of course would explain it at least.
 
As far as I know. See their website. Audyssey exists as sub entity of SoundUnited with their own staff / development team, which frequently releases updates.

So yes there would have been opportunity to change the default setting for that wretched “BBC dip”.

Or does SoundUnited sell speakers with that flaw? That of course would explain it at least.

My understanding is that in many cases, such a "flaw" is/may be inherent in some crossover desigs that, from the speaker designers stand point, may or may not be intentional as it depends.. Audyssey certainly did it intentionally, and they may or may not have a valid (to some extent only..) argument to say while it may or may not be a flaw in speaker design, but it's not their "flaw".

Audyssey's official position on this appears to be the following (I highlighted "vast majority"):

"Midrange compensation is an intentional dip in the 2 kHz region where the vast majority of tweeter-to-midrange crossovers are. In that region the tweeter is at the low end of its range and the midrange at the high end of its range and the directivity of the speaker goes through major changes. We found that if that region is equalized to flat, the change in direct to reflected ratio that happens because of the directivity variations causes voices to sound harsh (among other things). So, we have this implemented in the Audyssey target curve. With MultEQ Pro you can choose to turn it off, but we don't recommend it. This notion was observed 40 years ago by BBC speaker designers in their studio monitors. They designed their speakers with this "BBC dip" intentionally in the speaker response."

Regardless, Audyssey should have make their MRC feature selectable from day one, that might have helped D+M, Onkyo, and others sales from way back.

Some brilliant crossover designers such as @Dennis Murphy would be able to shed some light on the validity of the MRC concept. Wish he would show up once in a while.
 
My understanding is that in many cases, such a "flaw" is/may be inherent in some crossover desigs that, from the speaker designers stand point, may or may not be intentional as it depends.. Audyssey certainly did it intentionally, and they may or may not have a valid (to some extent only..) argument to say while it may or may not be a flaw in speaker design, but it's not their "flaw".

Audyssey's official position on this appears to be the following (I highlighted "vast majority"):

"Midrange compensation is an intentional dip in the 2 kHz region where the vast majority of tweeter-to-midrange crossovers are. In that region the tweeter is at the low end of its range and the midrange at the high end of its range and the directivity of the speaker goes through major changes. We found that if that region is equalized to flat, the change in direct to reflected ratio that happens because of the directivity variations causes voices to sound harsh (among other things). So, we have this implemented in the Audyssey target curve. With MultEQ Pro you can choose to turn it off, but we don't recommend it. This notion was observed 40 years ago by BBC speaker designers in their studio monitors. They designed their speakers with this "BBC dip" intentionally in the speaker response."

Regardless, Audyssey should have make their MRC feature selectable from day one, that might have helped D+M, Onkyo, and others sales from way back.

Some brilliant crossover designers such as @Dennis Murphy would be able to shed some light on the validity of the MRC concept. Wish he would show up once in a while.
That’s their official position which however in modern speakers is not true anymore. Look at the speakers Amir and Erin test and recommend. I at least don’t remember a single one with that dip.

It only applies to those old “BBC speakers and their derivatives”
 
That’s their official position which however in modern speakers is not true anymore. Look at the speakers Amir and Erin test and recommend. I at least don’t remember a single one with that dip.

It only applies to those old “BBC speakers and their derivatives”

I know, but my point is, even way back when there might have been many more speakers that behaved with the dip, it would not always be too close to 2,000 Hz and/or not dip as much so they should have provided the feature as optional/selectable. The sad thing is, until the Editor App shows up, the only possible way to disable MRC is to use the expensive pro version that was also sold to professional installers iirc.
 
Thank you Dr. Toole, as always, for your wonderful educational message!



Yes, I always feel the same.

In this perspective, I usually do not like the "psychoacoustic smoothing" after the very short-time sine wave scanning, as I wrote here.

At least for me, my "cumulative white noise averaging method" (please refer to here and here) would much better fit with my actual listening sensations, and the so-given "house curves" are highly reproducible, less statistical fluctuations/deviations all over 20 Hz to 25 kHz. Furthermore, I can choose proper FFT size (as smoothing intensity) depending on the Fq zones of interest. Of course, the quality (QC assurance) of the "flat white noise" should be critical; I use the very strictly well prepared and QC-ed "flat white noise track" of "Sony Super Audio Check CD".

Let me share an example of such "house curve" measured by "cumulative white noise averaging method" in my next post bellow.
If properly done both swept tone and noise analysis should give identical answers. It is a choice. The principal difference is in the heating of the drivers in sustained tests at high sound levels - power compression. Low frequencies require longer averaging times.
 
I like my house curve. Same settings for music and movies. (Revel Peforma3 F208 + IB 2x18")

index.php
 
Favorite house curve?
Who can live with just one.....???

My real-time house curve controller :)
house curve contoller.JPG
 
OK. I thought about opening a rat's nest by doing this, because it is a simplification of what sometimes has to be done. Some of the points have been made earlier in this forum thread and elsewhere, but it might be useful to bring the key factors in the process to one place. The marketing of room EQ algorithms often presents the impression that all combinations of loudspeakers and rooms can be "fixed", "calibrated" or the like, by means of measurements, math and equalization. In reality, much of the "math" does not include the exceptionally complex, non-linear and occasionally capricious psychoacoustics of human listeners. A critical missing element is that humans adapt to circumstances, bringing our perceptions into acceptable territory. Loudspeakers reproduce sounds. Musicians produce sounds. Both do it in rooms. We don't feel the need to "equalize" - even if we could - the instruments and voices of live music. Two ears and a brain separate the sources from the venue, and adapt to aspects of what the environment contributes to the overall performance. The venues vary, and some are even not ideal, but we manage to appreciate the excellence of fine instruments and voices in most of them.

The special problem with sound reproducing systems is that flaws get superimposed on everything that is played through them. These monotonous colorations can sometimes be beyond the ability of humans to adapt, and they need to be identified and attenuated.

Therefore, the "right way" begins with choosing well designed, timbrally neutral, loudspeakers. If the loudspeakers exhibit audible resonances and/or frequency-dependent directivity issues, it is not likely that measurements in a room will reveal such problems and that equalization is capable of compensating for them. It is often the case that the solution is better loudspeakers. Fortunately these can be identified with good reliability from competently made anechoic measurements presented in a "spinorama" format, following the industry standard. Amir, on this site, makes such measurements and others can be found at www.spinorama.org.

This done, set them up in your room and make a steady-state frequency response measurement at the prime listening position - the stereo seat. We will be paying close attention to the frequencies below about 400-500 Hz, where adjacent boundary effects and room resonances are active. Because much of the bass in recordings is mono (all of it in LPs) drive both loudspeakers simultaneously to evaluate what is happening at low frequencies. Measure them individually to find out what is happening at frequencies above about 400 Hz. If you are using bass management and one or more subwoofers the process is the same, and of course all subs should be running simultaneously. Why? Because multiple sound sources couple energy to room resonances differently when they operate in unison.

You can repeat this at different seats to see how much seat-to-seat variation there is - often quite a lot. Averaging several of these curves is a common practice, making the curves look much smoother, but hiding some awkward realities at low frequencies. Superimposing the curves on one graph is a more useful display of what is happening in your setup. You can then choose which humps/peaks to attenuate, depending on which seats are affected. Remember, at this stage we are looking only at bass frequencies. Narrow dips, however deep, should be ignored. Broad dips can be filled in, but keep the EQ boosts below about 6 dB. Aim for a smoothish curve that is tilted slightly upward at lower frequencies.

The benefits of this exercise will apply only to the seat or seats exhibiting similar shaped curves. That is why multiple-sub methods have been developed aimed at reducing seat-to-seat variations so that one equalization can deliver improved bass to several listeners. These are discussed in detail in Chapter 8 in the 3rd edition of my book.

Above about 400-500 Hz the "early reflections" curve in the spinorama should be similar to what you have measured. If you have well designed loudspeakers the room curve might have some smallish ripples caused by acoustical interference between and among the direct and reflected sounds - these are not problems to two ears and a brain and equalization is the wrong method of addressing them if they were - that is an acoustics issue. Spatial averaging over several microphone locations tends to smooth the room curve at middle and high frequencies, thereby reducing the likelihood that an auto-EQ algorithm (or a person) might try to "fix" something that can't be fixed, or that doesn't need to be fixed. Remember, any EQ applied to a room curve modifies the direct sound, and it the the direct sound that is a key factor in determining sound quality. If you began with loudspeakers designed to have the desirable smooth and flat on-axis/listening window response, they will be degraded.

Finally, pay attention to the overall shape of the room curve. Usually, at least for conventional forward-firing loudspeakers, the room curve will tilt gently downward. If the shape deviates substantially from the early-reflections spinorama curve then one can suspect something is amiss in the acoustical treatment of the room. If listening confirms a problem, then one is free to try modifying the shape of the spectrum with broadband, low-Q, tone-control kinds of equalization. When listening to recordings we get into the circle-of confusion dilemma, where it is difficult to know where the problem lies: the playback system or the recording.

Don't worry about little ripples. When I see exceptionally smooth high-resolution room curves I strongly suspect that something wrong has been done. The measurement microphone is no substitute for two ears and a human brain. Happy landings!
I would take objection to the view that all music is created equal and then move over to sound reproduction. I can see from your book and other comments that you like symphonic music played in good music halls. Instruments for classical music, like violins, radiate a lot of their sound upwards. Unless one sits in venues like the Elbphilharmonie and others, one gets a low portion of direct sound a lot of indirect sound. To this day it is a miracle to me how this is reproduced faithfully with a frontal two speaker system. But this is not my music so I skip it (or listen to it on an x.y.4 system).
My favorite is instrumental jazz in small venues without amplification. Direct sound from trumpets mixes with half direct waves from a saxophone and the upward radiating piano and else. Sitting in row 10 and taking out an Iphone with an FFT app I see a beautiful downward sloping “house curve”. This is often spoiled by “sound engineers” that want to squeeze the assemble through two speakers. This is worst than listening to music in a cassette player. The speakers throw flat direct sound at me that halfway mixes with the indirect soundstage. I wonder if those engineers are responsible for recordings too - what an outlook!
Hence, recreating the said original soundstage’s sloping curve in my house with a designed full frequency house curve might be a good idea. Most Rock music originates from speakers only, throwing direct sound around, again I have no idea how this should be treated at home. And then there is acoustic pop music a la Norah Jones that is similar to my jazz stage example. Wanting to say: It depends. Full frequency room eq recreating a jazz stage atmosphere of a 300 m2 room in my 30m2 hard walled home with art on the walls instead of foam might be not as bad an idea after all.
 
Does this mean the "Midrange Compensation" dip that Audyssey recommends around 2kHz is not a good idea for well measuring speakers because it is changing the speaker's characteristics above 500Hz? How about the "High Frequency Roll-off" that they recommend?
Your question suggests that you still haven’t fully grasped that, with or without the ‘dip’, it doesn’t matter, it’s still a bad idea, because the whole process is relying on doing EQ to achieve an in-room target curve above 500 Hz.
 
Back
Top Bottom