• 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!

Question about EQing speakers above 500 Hz

What lead you to the assumption that I was questioning whether you understand it or not?

The post was more written for the OP, than for you.
You quoted me so I understood it as an reply to me... But then fine :D
 
Hi there

I was going to answer your original question in the room correction thread, but I gather you opened this thread in order to deal with the question in its own thread. So I will answer here.

In short: people are saying "don't use EQ above 500 Hz" with specific reference to the case where a microphone is used to measure the summed-sound frequency response of loudspeakers playing in a room, where "summed-sound" means the sum of the direct sound coming from the speakers, plus all the reflected sounds from all room surfaces and room objects, including multiple reflections over time. This, in fact, is exactly the measurement that automated room correction algorithms take. Unfortunately, our ears don't work that way. Well, they kind of do work that way below 500 Hz (some say 300 Hz, but let's use 500 for now), but they definitely don't work that way above 500 Hz.

That is why the advice is centred on 500 Hz. Below 500 Hz, the summed-sound frequency response is a reasonable indicator of what we will think the system 'sounds like'. So it makes sense to allow room correction algorithms to apply EQ in that range. But above 500 Hz, what we think the system 'sounds like' is much more aligned with the frequency response of the direct sound, not the summed sound. (Direct sound being just the sound arriving at the ears straight from the speakers, and excludes any reflected sound.) Above 500 Hz we do also respond to the attributes of the reflected sounds, but to a lesser degree than to the direct sound. But the measurement that automated room correction algorithms take is not direct-sound, it is summed-sound. So it is considered unwise to decide on EQ adjustments above 500 Hz based on summed-sound measurements. We are looking at the wrong measurement for that purpose.

That is not to say we should never EQ our speakers above 500 Hz, which is where I think you are getting confused, since you see Amir and Erin recommending such adjustments. If the frequency response of the direct sound from our speakers is not very smooth and flat, then it is good advice to consider applying EQ above 500 Hz... but it needs to be based on looking at a measurement of the direct-sound frequency response. This is exactly what Amir and Erin are doing. They have the right tool to measure the direct sound.

For a more detailed treatise on this point, take a look at this paper by Dr Floyd E Toole: https://aes2.org/publications/elibrary-page/?id=17839

cheers
Thanks. This kinda gets to the heart of what was making me scratch my head.

Also, thanks to everyone for chiming in. This threads is starting to devolve a bit, and I certainly wasn't looking to poke the bear. It was just something I wanted to understand. I am auditioning speakers at the moment, and the truth is, it isn't easy. The more I learn on the subject, I am hopeful the better I can start to understand what I am hearing. For reference, I have been working on my systems, off and on for about 35 years. I still feel like a noob.
 
Thanks. This kinda gets to the heart of what was making me scratch my head.

Also, thanks to everyone for chiming in. This threads is starting to devolve a bit, and I certainly wasn't looking to poke the bear. It was just something I wanted to understand. I am auditioning speakers at the moment, and the truth is, it isn't easy.
Really :)

Then we compound ^all that^ with everyone doing post hoc rationalisation where they swear that they have the best speakers.
I am no different, and gravitate towards 1st order crossovers, and “time and phase” correct.
Others may use DIRAC to get there.

The Toole and Olive research work says that frequency response is the most important, but there is also dynamics, imaging etc.
None of it is easy, and it is layered in psychology, salesmanship, etc.

The more I learn on the subject, I am hopeful the better I can start to understand what I am hearing. For reference, I have been working on my systems, off and on for about 35 years. I still feel like a noob.

There an Harmon app called something like, “How to listen”… from Sean Olive.

 
Interesting outcome (or not :facepalm: ) if you leave out 8% other it is close to 60% Schröder an 40% full range.
From a pure technical point of view i acspected something close to 90-10 regarding theory specific ask in a ASR group.

I voted full range on that thread, but what isn't captured in the results is that I only use wide filters and when I correct over Schroeder I usually do some MMM checking also... If you have a dip at e.g. 1khz everywhere in the room there's no reason not to gently correct it IMO.
 
The Toole and Olive research work says that frequency response is the most important, but there is also dynamics, imaging etc.
None of it is easy, and it is layered in psychology, salesmanship, etc.
Funny you mention that. Half of my wondering is likely that I can hear, but I might not know what I am listening for in the frequency range. I can tell if something is too bright or sounds good or recessed, but the small nuances don't seem to make a big impact on me.

I am more taken with imaging and dynamics. I own JM Labs Cobalt 816S. I got a pair of Sonus Faber Lumina II Amatores from a friend to try. Imaging was better, but they sounded thin and anaemic, even with the subs. I then picked up a pair of Kef Q Concertos to try as well on some crazy Boxing Day sale. Dynamic, full and great imaging.

One of my pet peeves about my JM Labs was that I felt like I needed to keep my head in a vice to keep my center image. Any tilt of my head, and the image drifted. If the lights are off, I can't even visualize where the center image is. I have to turn the lights on to look at nothing to visualize the singer etc. Makes ZERO sense to me. However, with both the Sonus Faber and Kef speakers, I can close my eyes or be in darkness and I can visualize the center image.

It definitely is a combo of different elements.
 
There is predictability of EQ’ing above the transition frequency versus practical results.

You lose predictability when the on and off axis differ as you get into the higher frequencies which are naturally more directional. Overly aggressive EQ can overlook the fact that simple moving your head a bit to the left or right can alter the correction *and* in many cases, EQ may come with phase compromised.

On the other hand, using something like “tilt” is very accepted by science. How does tilt differ from EQ above the transition frequency? The Q-factor and amount of tilt.

Last, how good are your ears from day to day? How accurate is your calibrated microphone from day to day?

Lots of questions, no clear answers even with science. I prefer full range correction but I correct gently with broad strokes rather than trying to correct ever tiny peak or valley.
 
Ufff you EQ what you can. You first depending on the methodology and availability calculate and addresses first room length dependant peek with; PEQ, Vbass (convolution) or actual opposite firing subs. In near field you can do it even with simple placing the sub little more than 1m directly ahead of you. With that out of the way and it's harmonics you do measurement and start to EQ.
With PEQ you can't hit it directly when Q (up to 50 usually) factor stops being adequate, you can still shape it but you can't hit precise spot. Depending how actually it looks you can directly hit the problem up to 1~2 KHz.
In the highs dispersion becomes the problem so you should use the accustic treatment which is very efficient there and doesn't require thick panels. Then with much better focus and decay in line with what it should be you can do a impulse response inverted correction but be careful not to try to address deeps (+1~2 dB is still possible but don't go more). In the first 500 Hz accustic treatment is; expensive, thick and not efficient so EQ is a much better choice. If you try to do highs with convolution in a "bad" room where they are running wild the results wont be good to say at least.
You need to get mids to upper mids and lower highs as good as you can as that's where everything is happening anyway and you pay the most attention and work to getting bass into shape and remember we do hear different so make sure you use equal loudness compensation to psy level it around.
 
Or one says, “Maybe I’ll try to get a speaker that needs less <or zero> EQ.”

The pursuit cam says, “No EQ, and no active crossovers.”
And the other extreme is powered/active speakers with amps for each driver.
The physics and our hearing say otherwise! So will you listen them or that other guy?
You do benefit greatly from self filters and digital crossover especially when it comes to sub's as you gain possibilities you otherwise wouldn't have it's just that it's not easy thing to do proper and probably as that not exactly for everyone. I don't want to go in depth in hire about either equal loudness or subs crossovers as I did many times already.
 
While physics explains / predicts behaviors and comprehensive speaker measurements are a very good indicator of how a speaker behaves, you don't always have access to them.
When both listening impressions and actual in-room measurements in REW (FR, impulse, waterfall,...) show improvements with full band RC compared to limited one, why not use it :) ;)
And if not, just limit it to Schroeder limit frequency.
Let's being pragmatic and test, with always the science as base.
Personally, Dirac always did a very good job full band for me (while with manual EQ, I never made it correct full band). I don't rely on my impressions but on REW measurements.
 
I always supply my friends with this to compensate for my room while having conversations....:)

1735288999268.png
 
The physics and our hearing say otherwise! So will you listen them or that other guy?
You do benefit greatly from self filters and digital crossover especially when it comes to sub's as you gain possibilities you otherwise wouldn't have it's just that it's not easy thing to do proper and probably as that not exactly for everyone. I don't want to go in depth in hire about either equal loudness or subs crossovers as I did many times already.
The purist camp is a real thing.

If we want a smooth handover between drivers then that sort defines the slope as being shallower.
The one thing that an active XO can give, is a much better grip on the drivers, than trying to shake the driver through a bunch of inductors and caps in the XO.

If subs were easy, then everyone would have them that do not overwhelm the music… but that is not the case.
It seems like it is difficult to do a sub where it blends seamlessly.
 
The purist camp is a real thing.

If we want a smooth handover between drivers then that sort defines the slope as being shallower.
The one thing that an active XO can give, is a much better grip on the drivers, than trying to shake the driver through a bunch of inductors and caps in the XO.

If subs were easy, then everyone would have them that do not overwhelm the music… but that is not the case.
It seems like it is difficult to do a sub where it blends seamlessly.
There is correlation in between the physical capacity and the knee for the equal loudness while you want to contain it on that exact driver. In that regard less than one procent does it right.
On another plane we can argue how the bottom low to sub extension should go and that would be mocking purists for the purpose. In this realm physics rule. Best you can is to adapt (and so do speakers and hearing). You can cut and to extend authend the driver output to the driver FS or under the ports output and you won't be able to do that with passive crossovers (you can but much higher with capacitor). Well having drawn math model right in front of you and working directly on it not costing you more if you add couple more steps certainly helps a lot. Decently done passive crossovers and on the right places of course are perfectly fine with me given how much work it takes to do them on something not hard coded as it will get messed up at some point anyway. It's really a make or break thing (crossover).
 
Thanks. This kinda gets to the heart of what was making me scratch my head.

Also, thanks to everyone for chiming in. This threads is starting to devolve a bit, and I certainly wasn't looking to poke the bear. It was just something I wanted to understand. I am auditioning speakers at the moment, and the truth is, it isn't easy. The more I learn on the subject, I am hopeful the better I can start to understand what I am hearing. For reference, I have been working on my systems, off and on for about 35 years. I still feel like a noob.
My 2 cents:

If your speakers have a flat on-axis response then toe in the speakers (so you do listen on-axis) and eq the room below 500 Hz.

If your speakers do not have a flat on-axis response then toe in the speakers (so you do listen on-axis) and eq them wide band (room and speaker) according to a valid on-axis frequency response measurement.

This will result in your personal house curve, which must not be equed against any "official" target curve. A house curve is the result of speaker, room and EQ and not the target.

But if you are not yet satisfied afterwards (maybe the off-axis response is not smooth) you can add additional EQ for personal preference. Nothing wrong with that.
 
I think correcting the impulse response (in the time domain) is useful as long as the distorting effect of the room is predominant.
As the wavelengths are reduced, therefore as the frequency increases, it becomes increasingly impossible and senseless to establish a valid correction on a non-infinitesimal volume.
Then remains the fact that whoever mixed the audio track did it with an equally non-linear system, so aiming too much at the nominal signal makes no sense.
However, given the perceptive prevalence of direct sound to the increase in frequency, it makes sense to purely correct in the frequency domain with a low Q equalizer.
At least, I have always found it useful and beneficial to balance the frequency response up to 20kHz or in any case up to the subjectively audible limit, especially to manage reflective surface and speaker directivity combination in the 3 to 8 kHz area.
 
They are wrong. I have used Dirac Live on four pairs of wildly different speakers (+ one sub) in my room – sometimes with additional correction with CamillaDSP. All speakers obviously benefitted from correction up to ~500 Hz compared to no correction and all speakers have benefitted to a larger or lesser degree from correction up to ~20 kHz. And even though I generally use a 'tightly focused' correction in Dirac the tonal improvements are apparent even far outside the sweet spot ...
my sentiments excatly.
 
I would like to see comparisons between the EQ adjustments Dirac make and properly made gated measurements of the same loudspeakers. There should be at least someone here who use Dirac and also knows how to set up measurements for a gated response in REW.
 
I would like to see comparisons between the EQ adjustments Dirac make and properly made gated measurements of the same loudspeakers. There should be at least someone here who use Dirac and also knows how to set up measurements for a gated response in REW.
That would be nice.
However, I think it would be even better to separate and observe not only the direct sound but also the reflection range through simple gating measurements. (Of course, the directional characteristics of a single microphone are different from the way we hear, or even binaural recordings.)
We listen to a combination of direct sound and reflections, and no matter how the direct sound changes, we also hear the effects of unpredictable reflections influenced by those changes.


Rather than the idea that "you shouldn't EQ above 500Hz," it seems more fitting to the thread's discussion to say that the results of EQing to match some general target curve are unpredictable.
As Dr. Toole often says, "The room curve is not a target but a result."
 
Last edited:
That would be nice.
However, I think it would be even better to separate and observe not only the direct sound but also the reflection range through simple gating measurements. (Of course, the directional characteristics of a single microphone are different from the way we hear, or even binaural recordings.)
We listen to a combination of direct sound and reflections, and no matter how the direct sound changes, we also hear the effects of unpredictable reflections influenced by those changes.


Rather than the idea that "you shouldn't EQ above 500Hz," it seems more fitting to the thread's discussion to say that the results of EQing to match some general target curve are unpredictable.
As Dr. Toole often says, "The room curve is not a target but a result."

Yes, it seems like many people get confused by the statement; “don’t EQ above 500 Hz”. But that only applies when you only have the in-room measurement to go by, but if you also have gated measurements it the anechoic data you can base the EQ adjustments on those measurements for the range above 500 Hz, and use the in-room response measurements for the range below 500 Hz.


My suggestion for an easy solution to get sufficient data is to make one MMM measurement from the listening position, which you base the EQ adjustments for the range below 300 to 500 Hz. And another MMM measurement at a short distance of 50 to 80 centimeters at the height of the loudspeaker’s acoustic center, as this measurement will pretty much mirror the result of a gated measurement, at least good enough as can be seen in the comparison I made below and it follows the little loss of energy from 4.5 kHz to 6 kHz.

1735376052412.jpeg
 
I voted full range on that thread, but what isn't captured in the results is that I only use wide filters and when I correct over Schroeder I usually do some MMM checking also... If you have a dip at e.g. 1khz everywhere in the room there's no reason not to gently correct it IMO.
Additionally, for instance, Dirac takes multiple measurements in various positions -back and forth, high and low, etc. The results would be something similar to a MMM in the same area.
Dirac likely avoids excessive high-Q corrections when its algorithms detect reflections, and I have confidence in its proprietary methods.

I may be digressing a bit, but ultimately, full-range correction works best for me in my room with my setup. :)
 
Back
Top Bottom