DSP: EQ’ing Full Range or Not – Testing the Ears

[Sonnie Parker]

There are a few well respected individuals out there that tell us not to EQ above 500Hz, that we may be doing more harm than good when we do. I have read where several have written something along the lines of once we see the nice and pretty and flat response, our bias kicks in and we automatically think it sounds better. I suspect that’s probably true for the most part, as it’s hard for our brains to be honest with us about all things audio. When we spend mega bucks on amps, processors, and speakers, that same bias is likely going to kick in and cause us to believe it sounds better. Afterall, who wants to spend $10,000 and turn around and say “I made a mistake”… especially if they cannot send it back? There is too much pride going on for most people to admit a mistake. Besides, it is going to sound better because you did all your research, read great reviews, read numerous forum posts speaking highly of it, your friends have been encouraging you, and again, you have spent thousands of dollars to get it. How could it possibly NOT sound better?

If we use the same thought process behind not eq’ing above 500Hz because once we see the nice and pretty response we will prefer it, we may as well never read another thing about any equipment we are considering purchasing because once we read nice and pretty things about it, if we buy it, we will prefer it. Same logic IMO.

We decided to do some unofficial blind testing during our recent speaker evaluation to listen to partial eq vs full eq. Wayne, Dennis, and I were present during our evaluation event. There was nothing sophisticated about our DSP testing, we just did it like we wanted to do it. We are not trying to be scientific or please anyone but ourselves, so if it is not to your liking, you are more than welcome to ignore it… we would actually prefer you did. We will share what we learned with those that might be interested and find the information useful. Whether it proves anything to anyone else is not something we care about… but it proved to us that we are certainly not ready to throw the towel in on eq’ing above 500Hz based on what a few people believe, regardless of their credentials. They say our brain and two ears hear differently than the mic… well then, let your brain and two ears tell you whether you like partial eq or full eq better. However, the only caveat is that you need to have someone else do the eq’ing for you and not allow you to know which is which, then your brain and two ears can actually be honest for you to make an intelligent decision.

During our event we setup 4 different speakers for evaluation. With each speaker I personally setup three different Dirac Live target curves. We had three presets… one with Dirac eq’ing partial correction without subs, one eq’ing full correction without subs, and then one with subs and random partial or full correction. Dirac could also be turned off on either preset with the push of one button on the remote. Dennis and Wayne left the room while I did the setup and had no clue as to what preset was what. It was completely random on my part as to how I setup the presets, and it was not always the same for both guys, and they alternated who listened first. I level matched them as close as possible, attempting to keep the full correction along the average response of the partial correction, and keeping the subs level with the lowest frequency of the main speakers, except for the MartinLogan 15A's... where there was no need to compare subs since they have four 12-inch subs with four 500-watt (1000 watts dynamic) amps, although we did listen and compare with subs (it was not level matched). Below are examples. Wayne will be posting more graphs with each speaker thread; these are merely examples to show you how I levelized the presets.

With every speaker Dennis could tell the difference between the three presets, although it was not drastic on any of them. He was always able to identify which was corrected full and which was corrected partial, as well as whether preset 1 or 2 matched 3. In 3 of the 4 he preferred the full correction (before I identified which was full)… only in one instance did he state he thought the partial correction sounded slightly better. However, with all four speakers he did state he could likely live with either correction. He stated he noticed a slight edge on several of the partially corrected presets, and with some songs he liked it, and with other songs not so much, although again, it was not drastic, and he could live with either, yet still preferring the fully corrected if he had to pick. There was one instance that I noticed Dennis shining the flashlight over at one of the subs to see if it was playing. He had a difficult time determining which preset the subs were on.

Wayne verified he was able to tell the difference in the presets with three of the four speakers, and on one he said he really could not tell much of a difference. He liked both presets between full and partial correction, and said he could live with either one, and furthermore he did not feel either one was adversely affecting the sound in any way whatsoever. With the MartinLogan speakers he did actually prefer full correction, but it was not a deal breaker.

As for me… I knew which presets were what, yet I was NOT able to clearly identify between the full and partial corrections… they were very close for me. I was able to identify when the subs were playing on a couple of songs, but not all, except for the A7X speakers. I felt the A7X needed subs on most all the songs I was listening to, and it was noticeable enough to cause me to prefer the subs.

It could be that my room setup accommodates full correction better than others. Quite possibly due to the treatments and the close proximity of the speakers to the listening position, along with the speakers being pulled on out into the room more than most people are able to accommodate. It is hard to say, but either way, we could not verify that full correction as doing any harm to the sound, and in fact was slightly more preferred for the most part.

ADDITION: We should also note that while we evaluated 4 very different speakers here, it will ultimately all come down to the speaker, your room and your ears as to whether EQ'ing full range will help.

While Wayne and Dennis may give more precise details about what they heard, we are not going to further justify what we did, or argue with anyone that is hard-headed and adamant about not eq’ing full range, and/or about something that perhaps they haven’t even tested themselves, or maybe only tested a time or two, and hasn’t done blind testing for themselves, as that is truly the only way for you to know if it will work for you in your room. If you want to remain as you are with what you do, so be it… if you want to experiment and try it for yourself and attempt to be honest with yourself, you may want to try it. We fully realize a man’s opinion changed against his will is of the same opinion still.

 

[Wayne A. Pflughaupt]

Great post!

...but it proved to us that we are certainly not ready to throw the towel in on eq’ing above 500Hz based on what a few people believe, regardless of their credentials.

As you may know, I’ve been EQing full range with high-end equalizers since the mid-1990s. I found that it could yield audible improvements, but it was also really easy to make a mess of things.

Therefore I never put much stock in the full-range nay-sayers for at least a few reasons.

For starters, they seldom (if ever) revealed what equalizers they were using. I mean, they could have been talking about a stereo seven-band Radio Shack for all we knew.

Even assuming the use of quality 1/3 octave or parametric EQs, it’s still anyone’s guess if they had a clue how to properly use them.

Some, I got the impression, hadn't actually attempted full-range themselves, but were merely parroting the failed experiments of others.

That said, I've never used them myself, but I get the impression from reading Wayne’s past evaluations on Audyssey and Dirac that modern high-end DSP systems are able to perform ultra-fine filtering that would be crazy to attempt manually. It pretty much takes the “idiot” factor out of the equation. So, I’m not sure what leg the nay-sayers are standing on these days.

Regards,
Wayne

 

[AJ Soundfield]

There are a few well respected individuals out there that tell us not to EQ above 500Hz

Don't know anyone who says that, that I would respect.
Now if they said, don't EQ a speaker that doesn't need EQ > 500hz, such as a speaker with very smooth on and off axis native response, then yes. I would respect that view based on all the perceptual science supporting it, vs zero against.
Then if they said if EQing, only EQ the quasi/anechoic polar response along the design axis, that is, very judicious EQ of the on axis when there is a corresponding linear distortion off axis, >500hz, I would respect that also.
Such nuances are rather critical. Much like manual vs "auto" EQ.

cheers

 

[AJ Soundfield]

Unfortunately forums folks will not be able to understand the implications of this type evidence, vs "belief"

 

[pratul]

My processor Anthem AVM70, has ARC which by default EQs till 5KHz. After reading posts that said EQing past 500Hz was bad, I tried to limit ARC to 500Hz. The difference was very noticeable. I then tried to EQ up to 10KHz and didn't hear much of a difference between the 5KHz and 10KHz limits. I currently use the default 5KHz limit and am happy with the sound.

 

[skid00]

I am running Carver Amazing dipoles, in an 18' x 36' cathedral ceiling great room.

The Carvers have a 7 dB peak starting at 1 kHz, and ending at 3 kHz. They are flat from 3kHz to 12 kHz. This flat area is 8 dB lower than 30 Hz to 500 Hz, which is mostly flat except for an 80 Hz to 90 Hz drop of 9 dB.

I have measured the Carvers from 0* to 90*, and (with the rear reflection acknowledged) they track almost exactly the same out to 6 kHz. There are high-Q dips (probably from the rear wave cancelling the front wave) at 100 Hz, but a strong 10 dB boost at 30 Hz.

I'm aware of the argument that off-axis sound cannot be effectively EQ'd, and that EQ can greatly increase distortion.

I have gone through dozens of EQ curves, most hand-adjusted from REW-calculated base curves while listening to my favorite test tracks. In my room, with my speakers, treble EQ is absolutely better sounding. I am running a flat curve (+/- 3 dB) from 100 Hz to 17 kHz (old ears don't need a treble drop-off), and a 7 dB boost starting at 100 Hz and peaking at 18 Hz.

Given how easy it is to create several matched-level EQ curves, and using eqAPO to quickly switch between, any audiophile can easily decide for him/herself if there is a benefit.

I'd also like to point out that with hand adjustment to deep base, with amplification and a woofer section that can handle the power, I went from a 30 Hz knee, down to an 18 Hz knee. And that is HUGELY better, with an astonishingly large number of songs (pop female vocals have <30 Hz bass?! - YES).

 

[AJ Soundfield]

The speakers to confirm (not "test") this with were the one speaker not used. The Revels.
Of course 3 men >50 yrs old preferring a significant "speaker" boost >10kHz is hardly news either.
The measurements shown appear to be "room", not "speaker".

 

[tesseract]

The speakers to confirm (not "test") this with were the one speaker not used. The Revels.
Of course 3 men >50 yrs old preferring a significant "speaker" boost >10kHz is hardly news either.
The measurements shown appear to be "room", not "speaker".

We did use the Revels for the evaluation, spent more time with that pair than any other. Also, Dirac did not boost the EQ to unnatural levels.

 

[AJ Soundfield]

We did use the Revels for the evaluation, spent more time with that pair than any other.

Any Revel measurements at LP and at speaker with EQ?

Also, Dirac did not boost the EQ to unnatural levels.

If those are LP measurements, there would be more boost at speaker > 10k. Unfortunately no "speaker" measurements to show what the EQ did to native response. But there's certainly enough shown, to be audible....and preferred.

cheers

 

[Sonnie Parker]

I should have known to post the Revels in the examples... although this will all be posted in the eval thread for that speaker, as mentioned above. No harm in posting the three corrected curves here too.

 

[Sonnie Parker]

I believe the F328's were very close to on-axis, as were the 210RT's... we had them spread out pretty well. Probably still 2.5-3 meters from MLP. The 15A's are closer for my preferences.

Either way... the guys listened... not a lot of differences, yet still preferred fully eq'd over partial. Repeat not a LOT of differences... per all of us. The differences actually took a little edge off, as Dennis described... perhaps relaxed the highs a little, but again, repeat repeat repeat, it was NOT a huge difference, only slight differences. Also... no one could comment on what they heard... they wrote it down and then after we all listened, we compared notes.

 

[JStewart]

Really interesting how the corrected and uncorrected responses track so closely in each example.
Must say something good about the room and setup. Not surprising looking at the responses there where not a lot of differences.
Also clear that great care was taken with mic position repeatability.
Nice work.
Can’t wait to read the rest of the story.

 

[tesseract]

@AJ Soundfield , we can see there is more energy above the room transition FR before Dirac/EQ. This roughness was identified and called out.

Yes, there is a boost up high, well above the 10 kHz point you have focused on. Let's agree that there is very little energy at that level AND that we are not hearing that particular boost/correction.

 

[Sonnie Parker]

Any Revel measurements at LP and at speaker with EQ?

If those are LP measurements, there would be more boost at speaker > 10k. Unfortunately no "speaker" measurements to show what the EQ did to native response. But there's certainly enough shown, to be audible....and preferred.

cheers

I can set the Revels back up and measure the uncorrected and corrected response at the speaker, 1 meter and 2 meter. I still have the projects saved. It won't change what we all heard, and is not necessarily an indication of what we actually heard either, because that is not where our two ears and brain were located at listening, but it might be interesting to look at.

 

[Sonnie Parker]

Here is one of the MartinLogan 15A full vs partial corrections. I did not include the sub in this one as it was not level matched. However, the comparison to the sub was of less consequence and concern with the 15A's, as they have four 12" subs with 500 watts each (1000 watts dynamic), although we did at one time level match them just for the sake of doing it. The main idea here was to differentiate between full and partial corrections. We also had two different sessions with these, two different target curves as Dennis wanted to hear them with a little less mid-bass, which he preferred.

I have gone back to post #1 and added the MartinLogan and Revel graphs so that all 4 are together.

 

[AJ Soundfield]

@AJ Soundfield , we can see there is more energy above the room transition FR before Dirac/EQ. This roughness was identified and called out.

Yes, there is a boost up high, well above the 10 kHz point you have focused on. Let's agree that there is very little energy at that level AND that we are not hearing that particular boost/correction.View attachment 43377

Dennis, my comments pertain to the graphs (Adam, JTR) that were posted prior to, not after my comments, like the Revels you show there.
That shows a treble cut, but again, was posted after my comments.

 

[AJ Soundfield]

I can set the Revels back up and measure the uncorrected and corrected response at the speaker, 1 meter and 2 meter. I still have the projects saved.

That would be swell.

It won't change what we all heard, and is not necessarily an indication of what we actually heard either, because that is not where our two ears and brain were located at listening, but it might be interesting to look at.

Well, seeing that my previous comments were about the Adam/JTR data/listening impressions, not the not yet posted Revels...

 

[Sonnie Parker]

Well, seeing that my previous comments were about the Adam/JTR data/listening impressions, not the not yet posted Revels...

You referenced not having the Revel graph... that comment would have to be in reference to the Revels... simply trying to appease you.

My comments apply to all of them.

 

[tesseract]

Of course 3 men >50 yrs old preferring a significant "speaker" boost >10kHz is hardly news either.
The measurements shown appear to be "room", not "speaker".

Dennis, my comments pertain to the graphs (Adam, JTR) that were posted prior to, not after my comments, like the Revels you show there.
That shows a treble cut, but again, was posted after my comments.

It is likely that most people of any given age would prefer the corrected response over the uncorrected. There is no "Old Guy Hearing Acuity EQ Compensation" setting on Dirac.

 

[ddude003]

I would suggest that the 690 sq in XStat™ CLS™ electrostatic transducer x 2 would sound better than most 1 in (0.79 sq in) tweeters x 2 and ~22 sq in mid range speakers x2 in a box... Just say'in...

 

[ddude003]

Unfortunately forums folks will not be able to understand the implications of this type evidence, vs "belief"

That sounds like a conspiracy theory... Possibly an alternate truth...

"forums folks"??? Is that like Frank Zappa's Po-jama people???

 

[AJ Soundfield]

You referenced not having the Revel graph

We now have the Revel LP "corrected" and uncorrected. Now all we need is the "speaker" data. 1.5m +/-, 0,45 off axis uncorrected and corrected to see what is "correct". I think by now it's pretty clear you didn't do a (c) like the "experts" did above. That's ok, (a) and (b) will suffice. I've found anechoic data for that model so we'll see.
Thanks

 

[AJ Soundfield]

It is likely that most people of any given age would prefer the corrected response over the uncorrected.

I wouldn't project that from a sample of 3 people using 3-4 speakers in Sonnie's room. It certainly appears under such conditions the EQ'd speakers were preferred. With no reference yet, I'll say "EQ'd", not "corrected".

There is no "Old Guy Hearing Acuity EQ Compensation" setting on Dirac.

With no "speaker" responses, the first 2 graphs posted certainly suggested so.

 

[tesseract]

I wouldn't project that from a sample of 3 people using 3-4 speakers in Sonnie's room. It certainly appears under such conditions the EQ'd speakers were preferred. With no reference yet, I'll say "EQ'd", not "corrected".


With no "speaker" responses, the first 2 graphs posted certainly suggested so.

I'll say "corrected", because Dirac does much more than EQ.

If what you are suggesting about hearing acuity were true, we'd see an unnatural rise above 10kHz at the LP, not the corrected flat response we see with each speaker. Dirac doesn't give us a v-shape at the LP.

 

[Sonnie Parker]

The point being... whatever Dirac did, it "improved" the sound at least slightly enough for preference, but it didn't kick the other option to the curb either.

We certainly aren't talking huge waves here, but preference does matter, even if it's minor. There are those that like to seek out every last drop of improvement.