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

[MediumRare]

Let me quote this
magnitude response is important but impulse response is as well... as is phase coherence between speaker pairs for better imaging.

Unless you're correcting by more than 10 ms, even 20 ms, you cannot hear any difference in music. If you have research that demonstrates otherwise, please share.

 

[dsnyder0cnn]

Unless you're correcting by more than 10 ms, even 20 ms, you cannot hear any difference in music. If you have research that demonstrates otherwise, please share.

I'm sure this is true…and 20 ms is very long for this sort of correction. But, I will say that the step response looks so pretty after correction. :-D

 

[Flak]

Unless you're correcting by more than 10 ms, even 20 ms, you cannot hear any difference in music. If you have research that demonstrates otherwise, please share.

What happens to the stereo image if you switch the polarity of one of the speakers?
Magnitude response and impulse response are discussed here:

 

[teyhyrh4r]

psychoacoustic smoothing would show the audible diferences better.
also fdw-windowing to compare the direct-ish sound

 

[MediumRare]

What happens to the stereo image if you switch the polarity of one of the speakers?
Magnitude response and impulse response are discussed here:

180 degree polarity swap in one speaker across all frequencies is a wholly different thing, of course.

 

[AJ Soundfield]

Aside from your personal attack on Amir

It was a joke. Lighten up Francis.

 

[AJ Soundfield]

Right now I am actually at 2.2 meters, after moving my chair forward a bit.

That's still pretty close to near field for a tower multi-driver speaker. That is precisely why the NRC measures at 2m, so that multi spaced vertical drivers across 1m +/- such as a tower speaker like the Revel, can properly integrate, along with baffle diffraction effects. Studiophile type folks unfamiliar with physics and speaker design can't know this, but that's why.
Looking forward to your "speaker" measurements to compare to the LP EQing.

 

[MediumRare]

It was a joke. Lighten up Francis.


Once again, the Klippel, Amir and yourself would have no clue how to interpret the total direct/indirect soundfield, since these are not direct radiators. This is prior to any room reflections whatsoever. Stay in your lane. My post you quoted has zero to do with EQ.

You say it’s a joke, yet you persist in insulting Amir and now me as well. You have zero idea who I am or my credentials, yet you insult me as having no idea about soundfields. I smell a troll.

 

[AJ Soundfield]

I smell a troll.

Then take a shower.

 

[AJ Soundfield]

psychoacoustic smoothing would show the audible diferences better.
also fdw-windowing to compare the direct-ish sound

Yes, an original vs EQ'd would be helpful. Bottom line is that all 3 listeners preferred the EQ'd version in Sonnies somewhat unusual setup, a largish room where one sits in a highly direct location. I wouldn't automatically transpose those results to a more typical setup with a greater amount of reflected to direct ratio.
Would still be interesting to see what he measured off speaker. These are direct plane wave radiators. Pretty straightforward.

cheers

 

[markus]

Yes, an original vs EQ'd would be helpful.

Done: https://www.avsforum.com/threads/the-official-dirac-live-thread.3137722/post-60888124

 

[AJ Soundfield]

Done: https://www.avsforum.com/threads/the-official-dirac-live-thread.3137722/post-60888124

Thanks Markus, but not quite. I'm hoping so see the "speaker" measurement with the EQ applied. The measurements below are "LP", albeit fairly close at 2.2m. I like your 1.5m distance, will work here for >500hz integration.
Note that at LP, (above 500hz), for example, the omni pressure mic measured some deviations >1kHz. The "corrected" response is also at LP and is now audiophile/studiophile eye friendly flattened. What we dont have is the "corrected" response at the speaker. Your example was a speaker no sane person would disagree needs "correction". Here we have a speaker that presumably needed little if any. We do have factory anechoic data for it. They differ from what is shown here.

 

[markus]

Thanks Markus, but not quite. I'm hoping so see the "speaker" measurement with the EQ applied. The measurements below are "LP", albeit fairly close at 2.2m. I like your 1.5m distance, will work here for >500hz integration.
Note that at LP, (above 500hz), for example, the omni pressure mic measured some deviations >1kHz. The "corrected" response is also at LP and is now audiophile/studiophile eye friendly flattened. What we dont have is the "corrected" response at the speaker. Your example was a speaker no sane person would disagree needs "correction". Here we have a speaker that presumably needed little if any. We do have factory anechoic data for it. They differ from what is shown here.
View attachment 43558

I did my measurements just to establish if DL is capable of improving the anechoic speaker response at all. Looks like it is. Now the question is if it does so under all conditions of interest. The procedure should always be ...
a) run DL from MLP in-room
b) measure anechoic response with and without DL active (preferably do full polar plots)

 

[AJ Soundfield]

We seek the same answers

 

[markus]

We seek the same answers

Maybe @Erin is interested as he was talking about doing a follow up on Dirac anyway?

 

[ciotime]

I posted this article and these are some of the responses I got....

" if you have issues and fix them in one spot you are making the speaker inaccurate and if that issue doesn't exist in another you've just made that spot worse. If you have an issue that needs attention You have to measure all the seats and see what's going on. You have to understand how Eqing for issues that exist here negatively effect the seat over there where that issue doesn't exist.
So while their room was treated and maybe they tested one seat or two and they used good speakers this doesn't hit on the reason we don't blindly do full range eq.
So yeah, sometimes you will need full range eq or it may sound great. This article doesn't cover any of them or discuss when you would or wouldn't nor explain what you're doing to the source and other seats."


"The writer of this thread has clearly no idea what the schroeder freq is, and what EQing above the schroeder does to the sound, and how it is a case to case basis for each speaker. "

 

[ddude003]

I think he was making a joke at Amir... as he's been mentioned a few times in reference to testing Harman speakers.

Clearly it is Klippel envy...

 

[Eric SVL]

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.

The fact that Audyssey doesn't let you see the measured response in the curve editor, so you are unable to trace the average level. That means Audyssey often sounds better when range limited because the default target curve isn't ideal for our speakers and our in-room response. So let's make sure we know what we are comparing:

Audyssey:
Range-limited vs default target curve

Dirac:
Range-limited vs custom curve

 

[Sonnie Parker]

I posted this article and these are some of the responses I got....

" if you have issues and fix them in one spot you are making the speaker inaccurate and if that issue doesn't exist in another you've just made that spot worse. If you have an issue that needs attention You have to measure all the seats and see what's going on. You have to understand how Eqing for issues that exist here negatively effect the seat over there where that issue doesn't exist.
So while their room was treated and maybe they tested one seat or two and they used good speakers this doesn't hit on the reason we don't blindly do full range eq.
So yeah, sometimes you will need full range eq or it may sound great. This article doesn't cover any of them or discuss when you would or wouldn't nor explain what you're doing to the source and other seats."


"The writer of this thread has clearly no idea what the schroeder freq is, and what EQing above the schroeder does to the sound, and how it is a case to case basis for each speaker. "

Doesn't surprise me in the least, as they have no clue what we were doing, nor did they completely read the article, otherwise they would not have been so ignorant in their response. We are only concerned with one seat. This is for music listening only, not movies. We are not trying to appease every seat in the room, only one, the MLP (main listening position). Either way, you'll never get every seat in the room corrected even for movies, which is why I've never concerned myself with other seats for movies. I worry about my seat... don't care about others... that's what I prefer. There is no reason to discuss other seats. I only have one seat for music (and movies), period.

The writer is full aware of what the transition frequency is (it's the whole point of the test), and if the dipwit that thinks I don't wants to come visit me and have a listen in my room and have a civil discussion about it all, get to know me better, then he can make a reasonable justified comment about it all. And if he read the article, he would "clearly" realize that we admit this is likely not only a "case to case basis for each speaker" (not those exact words, but by implication), but we also elude to the fact it is also a case to case basis for various rooms. We fully realize the room and the speaker (along with our brain and two ears) are all factors in whether you EQ above the transition frequency.

As I stated in the write-up, you'll have these kinds spattering nonsense, which is why we'd just as soon they ignore it, since they have no value to add.

 

[Sonnie Parker]

I thought I had saved the Dirac project, but apparently not, so I ran the entire setup again. It's not exact, but above 500Hz it's very close. I wasn't so concerned about the lower end.

The 2-meter measurements are all on-axis, while the MLP measurements will be slightly off-axis.

 

[Sonnie Parker]

And for kicks... here's the 15A.

 

[markus]

I thought I had saved the Dirac project, but apparently not, so I ran the entire setup again. It's not exact, but above 500Hz it's very close. I wasn't so concerned about the lower end.

The 2-meter measurements are all on-axis, while the MLP measurements will be slightly off-axis.

View attachment 43586

View attachment 43587

View attachment 43588

View attachment 43589

View attachment 43590

View attachment 43591

View attachment 43592

You need to compare the anechoic before/after responses not the steady-state responses.
Can you please post the .mdat so we can look at the data ourselves?

 

[teyhyrh4r]

Yes, an original vs EQ'd would be helpful. Bottom line is that all 3 listeners preferred the EQ'd version in Sonnies somewhat unusual setup, a largish room where one sits in a highly direct location. I wouldn't automatically transpose those results to a more typical setup with a greater amount of reflected to direct ratio.
Would still be interesting to see what he measured off speaker. These are direct plane wave radiators. Pretty straightforward.

cheers

if the system is direct sound dominant, it is obvious that a full correction will make it more balanced.
now a reverbant system can look messed up in the LP meassurement, while in reality having a fully balanced direct sound. in this situation you would make things worse with the full correction

 

[markus]

if the system is direct sound dominant, it is obvious that a full correction will make it more balanced.
now a reverbant system can look messed up in the LP meassurement, while in reality having a fully balanced direct sound. in this situation you would make things worse with the full correction

You assume that DL would not be able to derive a useful quasi-anechoic speaker response from its multiple measurement points. That assumption might be wrong.

 

[teyhyrh4r]

DL

I am sorry. what is DL?