Some comments from Floyd Toole about room curve targets, room EQ and more

[JoachimStrobel]

Hi, great collection of quotes - I found a few that I personally missed before.

Actually among everything I read, including the book I never understood what to do for people who have irregular listening space? All these recommendations to not touch speakers above transition frequency and avoid automatic calibration, but I have one speaker in a sort-of corner of a an irregular room, one in the middle of the wall, center in-between under a TV + two surrounds mounted at different distances in again strange corner conditions. Actually my situation is not much more extreme than Dr. Toole's own home theater today, but what could I use to set my system up if I don't want to spend lots of money on luxury equipment, don't want to have ugly professional rack-mount boxes and want the system to work with a simple blu-ray player as a source?

Is there a simple solution to a speaker setup problem for a consumer with an AVR? As bad as Audissey XT or similar systems are - they do get something right. Alternatively I could use quite basic manual EQ inside the AVR that may or may not be any better than Audissey when used with an external measurement system.

The funny thing is that this dillema stops me from buying really excellent speakers, as I know I will kill them by placement and any system that could fix the placement issues costs craaazy money (JBL Synthesis custom install, Trinnof, Genelec, even JBL Intonato 24 calibration).

Whenever I turn my Dirac processor on (Full Frequency), I hear two thing: The frequency response is different due to my selected room curve - and one could debate endlessly if it is for the better or the worse. And then the sound imaging is much different because of the subtle time alignment and level changes. And for sure, that is an improvement, small for stereo, great for Mch and superb for Mch with a distorted geometry. And this aspect is mostly forgotten when discussing the room curve. So for speaker placement, a full frequency DRC is great.
Being a fan of the NanoAVR, I need to say that is an engineering project to include it in a setup. What is needed is a multichannel pre-processor that takes all flavors of HDMI streams and a modular room correction, passing the signal to a Mch DAC (potentially w/o Hdmi) and then to the amps . There are bits and pieces available, but it seems that the all-in-one AVR kills any progress.

 

[Daverz]

Actually among everything I read, including the book I never understood what to do for people who have irregular listening space?

I wonder if using free-standing diffusors (mounted on casters, perhaps?) arranged to create symmetrical first reflection points might help. Might not fly with your interior decorator.

 

[SDX-LV]

Thanks everyone for suggestions but the "Multi-channel, 4K and HDMI requirement" still is a pain.

NanoAVR is tempting but relies on Dirac, costs more than my AVR with Audissey, was limited to old HDMI standard and 7.1 sound (not a long-term investment) plus is clearly not convenient for people with multiple sources.

"free-standing diffusors" - yea, not gonna happen in my livingroom At least I force my wife to not sit on a sofa against a wall when watching movies with me (also enlarges your TV btw!)

... What is needed is a multichannel pre-processor that takes all flavors of HDMI streams and a modular room correction, passing the signal to a Mch DAC (potentially w/o Hdmi) and then to the amps . There are bits and pieces available, but it seems that the all-in-one AVR kills any progress.

I can't agree more on this one. I wish to use powered monitors for most channels, but again it is quite expensive to get decent pre-amp outputs and avoid a steaming class-A amp sitting under a TV. As for equalization, it is tricky... At least there is hope that Dirac now learned to properly integrate multiple subwoofers. Then in most cases people can just disable speaker correction above 500Hz or less.

So no simple answer for irregular rooms for a consumer

 

[Thomas_A]

My experience with my quite big wide loudspeaker collection (from 70s till today) is similar, equalising loudspeakers with non continuous directivity to some kind of predefined target curve makes them often sound even worse, like for example filling a "BBC dip" which was there on purpose to compensate the decreasing directivity of a tweeter without waveguide.
So for loudspeakers with smooth directivity I only use above 300-500 Hz only anechoic data or room measurements from moving micro method which also reduces the influence of local disturbances like reflections. http://www.ohl.to/audio/downloads/MMM-moving-mic-measurement.pdf

Besides correcting for directivity problems, here have been several reasons in history for a dip in the region. What Linkwitz and BBC wrote about that was to compensate for some recordings made in larger orchestral halls with lots of diffuse sound, not to adjust for loudspeaker directivity problems.

https://www.linkwitzlab.com/models.htm#H
https://www.linkwitzlab.com/images/graphics/harwd.gif

Another reason for compensation is the stereo system errors, but that is another correction curve but where there should be peaking 1-2 kHz followed by dips in the 2-8 kHz region.

https://www.audioholics.com/room-ac...ons-human-adaptation/what-do-listeners-prefer

 

[digitalfrost]

Thanks everyone for suggestions but the "Multi-channel, 4K and HDMI requirement" still is a pain.

I think Acourate could probably do this. https://www.audiovero.de/en/acourate.php

 

[maty]

Again, KEF Q100 5.25 coaxials

https://www.audiosciencereview.com/...inimum-phase-vs-linear-phase.8762/post-225351

Measured by Zvu in a kind of sports center, with the speaker at a good height as measured in Audioholics in the open field.

[Big picture, KEF LS50] https://i.postimg.cc/sVm6Y2k9/IMG-2671.jpg

https://www.diyaudio.com/forums/mul...factory-cabinets-simulations-post5779288.html

Off-axis, At 30º, 40º and 50º

https://www.audiosciencereview.com/...inimum-phase-vs-linear-phase.8762/post-253826

From a good choice of the speaker, first the deficiencies caused by the cost savings and after the frequency response can be easily corrected. If listening is in the near field, everything is easier. But we need a good raw material, such as the fantastic KEF 5.25" coaxial Uni-Q drivers

And without spending a fortune!

 

[thewas]

Yes, I correct my LS50 also in the modal region according to my listening position and above to a smooth linear direct sound.

 

[maty]

Deleted so as not to distort the thread.

 

[thewas]

The polymere front of LS50 is acoustically more dead than usual MDF, but all these is off topic here.

 

[maty]

Deleted so as not to distort the thread.

 

[Robbo99999]

@thewas_ , thanks very much for the Toole links and the quotes from passages of his books, I'm making my way through them at the moment. I have a miniDSP based system on a 2 channel JBL speaker system for TV/Movie/(Music), and have been spending a lot of time EQ'ing them using REW with I admit not a massive amount of reading around the subject...well I've been reading bits here & there over the weeks as I've been EQ'ing them. I think I've been a bit too harsh with my EQ'ing re applying EQ to the whole curve rather than focussing on the bass, and I also didn't know about the advice that you shouldn't EQ at the crossover point between woofer and tweeter. Regarding that last point I have a question, when it's said that area shouldn't be EQ'd does that mean you shouldn't EQ out the dip at the crossover point, or does that even extend to not even being allowed to create a broad (low Q) filter over that general area that spans the crossover point with a view to changing the overall angle of the response in the high mid range/treble area?

In the past I've also been trying to EQ out sharp dips in the curve too, so I think I'll need to rethink that. From what I've read it seems like the main goal is not to create (at all costs) a perfectly smooth exact fit to the Harman Curve.

 

[thewas]

Above the transition frequency (around 300-500 Hz in usual rooms) as Toole says you mainly correct the loudspeaker, so if it has a problem there like for example in the mid/tweeter crossover that also appears in its anechoic angle measurements like for the example of the here measured Klipsch RP-600M it is not bad to correct it with an appropriate EQ.

 

[Robbo99999]

Above the transition frequency (around 300-500 Hz in usual rooms) as Toole says you mainly correct the loudspeaker, so if it has a problem there like for example in the mid/tweeter crossover that also appears in its anechoic angle measurements like for the example of the here measured Klipsch RP-600M it is not bad to correct it with an appropriate EQ.

Although didn't Toole say you shouldn't EQ out crossover dips? I'm definitely with you on understanding the transition frequency and the difference between correcting for the room below this, whereas above you should really only be correcting based on anechoic measurements of the speaker...but with the latter where does that meet with Toole saying you shouldn't EQ out the crossover dip?

I can't find any anechoic measurements of my JBL 308Mkii unfortunately, will measuring the speaker from say 6" with UMIK microphone offer any insights? (I won't be able to measure outside, speakers will be in my room).

 

[thewas]

Toole says not to EQ a crossover dip if it disappears under angles and thus at the sound power, so usually at loudspeakers with not smooth directivity. Do you want to EQ your JBL 308P MKII? If you want you can post your measurements, but I strongly doubt the 308 has a significant problem at its crossover frequency that should be corrected, especially since the MKI didn't have one.

 

[Robbo99999]

Cool, I'll do some measurements of my speakers from 3 close positions, within 30cm of centre position and post the average. I'll do that with and without a low shelf boost around 110Hz, as to enable a "Full Harman Curve" I think it needs the low shelf boost which I like for music listening, whereas for TV/Movie I prefer a "Half Harman Curve" that might fit better without the Low Shelf boost. I might then see if I can some pointers from you. I'll post it up this afternoon.

 

[Wayne A. Pflughaupt]

In the past I've also been trying to EQ out sharp dips in the curve too, so I think I'll need to rethink that. From what I've read it seems like the main goal is not to create (at all costs) a perfectly smooth exact fit to the Harman Curve.

Someone recently at AV Nirvana had a great quote for this situation: “I’ve had the feeling that my overzealous EQ attempts fixed the graphs but killed the sound.”

Regards,
Wayne A. Pflughaupt

 

[QMuse]

I have tried both kind of equalization on my system : speakers with flat anechoic response + bass correction, and full band equalization from the listening position.

The speakers are Neumann KH-120 in a reverberant room. They have a completely flat frequency response curve without correction.

View attachment 46588

Listening distance : 2.1 meters, decay time at 500 Hz : 0.45 seconds. Treble tilt set to -1.

It took a very long time, with trial and error, to properly setup the manual bass correction. Starting from two huge peaks at 54 and 69 Hz, with a big depression around 100 Hz, very strong correction, with unstable results, are needed. Here are 6 one-point sweep tone measurements made 30 cm around the listening position:

View attachment 46585
And the final correction (left / right) :

View attachment 46586

In the end, the result sounds perfect, although it measures oddly : the level decreases about 2 dB from 200 to 40 Hz instead of increasing. I suppose that this is was I preferred because the result is not smooth. Here is the result with both speakers, measured two times, two months apart.

View attachment 46589

Then, I tried a complete correction, made by Jean-Luc Ohl, sending him the measurements, and getting two impulse responses in return (with a lot of interesting graphs).

View attachment 46587

The result was worse : since the impulse response were about 6000 samples long, the bass peaks were no more corrected properly, and also the overall balance was too bright, because the target curve that was chosen was not steep enough.
It means that the direct sound was not flat anymore : there was a treble boost.

After comparing the measurements, Jean-Luc sent me another correction, that followed more closely the natural curve that I have got at the listening position when the direct sound of the speakers is flat.
It sounded better, but it was impossible to compare it with the setup with bass correction only, as the bass themselves were not properly corrected, and that was too distracting.

So I decided to build a mix of the two corrections: I added my own correction for low frequencies to Jean-Luc's automatic correction for the rest of the frequencies:

View attachment 46591

Now, the difference between this mixed correction and the pure correction C above is very small. It is difficult to tell which one sounds best, but I have a small preference for correction C, with no change in the direct sound of the speakers.

Comparing very carefully the frequecy responses of both corrections measured at the listening position, I realized that the main differences were a lack of accuracy in the 200 Hz range, and yet a slight change in the general balance, the green curve being overall lower in the 600-2000 Hz range, and higher in the 2000 - 8000 Hz range:

View attachment 46593

In conclusion, the obvious variations that I have between 1000 and 10000 Hz measured at the listening position were not a problem. The problems with automatic corrections never came from these unwanted corrections, but always from a general target curve that was not perfect.
The only perfect target curve, that sounds natural, is the one got with speakers equalized in anechoic conditions.
Looking at the last correction above, the two positive corrections at 2000 and 5500 Hz in the "correction Ohl" part are not heard as audible problem in themselves. The problem is that they are raising the average level of the whole 1000 - 20000 Hz range. They should be compensated by a lower overall level.

In fact that's the very problem that I've been facing from the beginning while equalizing the 35 - 600 Hz range : should I decrease the peaks or rather raise the shelves between the peaks ?
Note that I'm not saying "filling the dips" but "raising the shelves" (sorry for my bad english), as filling a narrow dip is usually very bad, while adding a general low shelf and decreasing the peaks accordingly is harmless.
In other words, where is the target supposed to be ??

So far my answers are: from 200 to 600 Hz, the correction added after the speakers are flat on-axis in anechoic conditions should be balanced between positive and negative, so that the average result in the 200 - 600 Hz range for the direct on-axis sound the speaker remains at the same level as the one above 600 Hz.
Below 200 Hz, I don't completely understand what's going on. It seems to me that my correction is still very imprefect, and that I prefer having this frequency range in the background, as it still sounds quite bad compared to the clean bass in headphones such as Sennheiser HD600 or Beyer DT-880 pro.

In my opinion range up to 100Hz is corrected well. Range up from 700Hz should anyhow be corrected based on near field gated mesaurement, here it looks ok to me.

It is the 100-700Hz part I'm finding problematic:

Red line is how it should be, but unfortunately variations from that line are severe. As most of them are above the line they will have large impact on balance with low bass and with HF as well. So, IMO opinion you should concentrate your EQ efforts in the 100-700Hz region and try to make it close to the red line. It should be relatively easy to do as small narrow dip at 140Hz can be left untouched while wider 250-400Hz dip is only 2.5dB deep and can be corrected with a single filter.

P.S. If you post this measurement in .mdat form I'd be happy to propose few PEQ filters made with REW that will improve the 100-700Hz range

 

[QMuse]

Thank you Joachim, here you can read more about that graph http://www.aes.org/e-lib/browse.cfm?elib=17839

Also thank you very much Pio2001 for your thorough post which confirms my past experiences, as said "correction" bases on listening position measurements of very neutral loudspeakers like your KH120 above modal region is usually counterproductive, so its better done on anechoic measurements.

About the lower region my experience is similar too, when using a neutral speaker like your Neumann the target curve should be somewhere in the middle of the peaks and dips as they are the influence of the room and local reflections but that depends also on the recording (some old recordings are quite bass shy as they were mixed with huge monitors without room EQ back then) and playback level (when listening with less than 85 dB a little bass boost may sound more neutral due to the loudness effect).

Funnily I have found that one specific Harman target curve fits in my room very close to the moving microphone listening area measured responses of my loudspeakers, so when correcting to it I also linearize the direct sound which is very convenient, as I don't need a 2 step correction as before.

Here for example my correction with it on my desktop speakers, the passive KEF LS50 which have very smooth horizontal and vertical directivity but not perfect frequency responses:

View attachment 46610

Here my measured windowed responses (so quasi anechoic) of them without and with above EQ:

View attachment 46613

As it can seen the quasi anechoic response is nicely linearised in the upper region to a approx. +- 1dB width and the little frequency response problems of the LS50 like its famous presence "hill" are corrected.

But as said this only works because of the smooth directivity and because that target curve fits my room response so well already before correction. Like Dr. Bruggemann of Acourate always says when people ask him about target curves, the best it to look at the difference to your current measured curves.

This is done "by the book" - congrats, I'm sure it sounds equally well as it looks!

 

[Thomas savage]

Please stay on topic in this thread , it's highly valuable so will be moderated and posts removed.

Cheers

 

[QMuse]

Someone recently at AV Nirvana had a great quote for this situation: “I’ve had the feeling that my overzealous EQ attempts fixed the graphs but killed the sound.”

Regards,
Wayne A. Pflughaupt

IMO one of the main reasons behind is because very often the final result is not checked with REW and to compare it with the listening impressions.

Quite rarely I have seen good looking control measurement in REW resulting in a poor sound.