Dirac looks into mixed and minimum phase issues, I am sure other software packages do too.
-
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!
EQ, Is too much of a good thing bad?
- Thread starter Ron Texas
- Start date
Manual PEQ's works better for me in the time domain than auto-generated biquads with Minidsp. And yes, too much EQ will not sound good. I recommend taking a few measurements around the LP both in height and width just to get a sense of the overall picture before hammering the response completely flat.
In theory you can get this kind of response if you use one point in space and try to hammer it flat;
I tried Dirac measured in only one spot. Same smoothing in both sets of lines, Minidsp was before and Dirac was on top of that. Insanely smooth lines, but sounded awful. The same is true if you try to equalise in Minidsp with Biquads in several channels to get a smooth frequency response.
I agree completely with Armin, above the transition-area, it's all about tone adjustements. As to differences between room correction systems, it's not black and white. Dirac relies heavily on good spacing between measuring points in order to sound good. Audiolense works with one point, but is so customizable that you can alter how and where it will correct and by how much. You can get Audiolense to sound insufferable, but you can also make it sound extremely good. Regardless, here's a comparison between standard default settings between the two;
Dirac - Red
Audiolense - Purple
In theory you can get this kind of response if you use one point in space and try to hammer it flat;
I tried Dirac measured in only one spot. Same smoothing in both sets of lines, Minidsp was before and Dirac was on top of that. Insanely smooth lines, but sounded awful. The same is true if you try to equalise in Minidsp with Biquads in several channels to get a smooth frequency response.
I agree completely with Armin, above the transition-area, it's all about tone adjustements. As to differences between room correction systems, it's not black and white. Dirac relies heavily on good spacing between measuring points in order to sound good. Audiolense works with one point, but is so customizable that you can alter how and where it will correct and by how much. You can get Audiolense to sound insufferable, but you can also make it sound extremely good. Regardless, here's a comparison between standard default settings between the two;
Dirac - Red
Audiolense - Purple
ernestcarl
Major Contributor
Interesting. Is there a before correction image for reference? I am assuming from the above that you have Dirac run full-range. Last time I used it on the neumann's, correction above 200-300Hz didn't sound right to my ears no matter what "house curve" was used -- then again, these monitors are already ultra-linear to begin with. I have had better results with just using the shelving switches already built-in the unit.
Sorry, I have so many measurements stored with insufficient description, so I don't know which ones. Here's an example of Audiolense vs only minidsp-correction in the bass and broad-band tilt from about 3 khz. It should be somewhat similar measurement from the ones above;
(Note that the fuzzy behavior in the high frequencies are due to stupid placement of the microphone stand)
(Note that the fuzzy behavior in the high frequencies are due to stupid placement of the microphone stand)
Here's an example, with "flat" as the target curve.
AcourateDRC creating the FIR and IIR filters for a MiniDSP OpenDRC-DI, 1/12th octave smoothing
Red = Raw
Black = Corrected
48Hz hole is phase cancellation in my asymmetrical room at the listening position, 220 and 440Hz is (probably) the bounce off the back wall - dipole speakers.
The FIR filter generated can move the phase (unsmoothed ):
The discontinuity, at 400Hz in this graph, is related to the crossover from panel to woofer at 180Hz - sometimes the jump is reported at different frequencies, but it is always there someplace.
Distortion is not adversely affected:
The high distortion levels reported around 48 and 220Hz are due to the lower SPL at those levels as mentioned above
Impulse Response:
Step Respnse:
A little rise before the step since the bass is moved relatively forward in time.
Group Delay (1/12th smoothing):
Last edited:
- Joined
- Feb 5, 2019
- Messages
- 3,688
- Likes
- 4,070
I can agree if you talk about room correction.
I don't agree if you talk about speaker correction.
Did you read the site NOAUDIOPHILE and see what the guy is doing to correct a lot of speakers? He basically attacks peaks only ABOVE 700Hz.
Many many years ago, my father’s high-end amplifier had separate treble and bass adjustments for left and right. When I bought my high-end gear in the 90’s he searched for the treble & bass adjustments, and after finding none he walked away (not from me, but from my equipment).
And now DSPs are creeping in again. A couple of thousand dollars later I have a DSP with 4 switchable frequency adjustments. My father would have had tears in his eyes. I probably too after I played his Quadro SQ discs through a Dolby matrix... . Circle of invention.
And now DSPs are creeping in again. A couple of thousand dollars later I have a DSP with 4 switchable frequency adjustments. My father would have had tears in his eyes. I probably too after I played his Quadro SQ discs through a Dolby matrix... . Circle of invention.
ernestcarl
Major Contributor
From the image it certainly looks like it did a pretty good job evening out the bass. I suppose as as long from your listening tests the overall sound is not aversly affected, then all well and good. Although I am curious: does Audiolense allow the user to see the filters it generates?
ernestcarl
Major Contributor
Ray, I don't think I've ever seen something like that negative group delay effect before -- after your Acourate correction. Although you did say it 'time corrected' the impulse response. Still, it looks rather a bit strange.
While I'm quite certain that you likely would have noticed much improvement in the bass... to you, were there any marked improvements, audibly, that is, correcting way above that?
The room is a rectangle on the right, and open on the left rear corner. that makes the left side both deeper and much wider than the right side.
At 48Hz the left and right bass is 180 degrees out of phase at the listening/measuring position.
So, looks weird there.
L, R, and Both speakers, corrected frequency response,
1/12 smoothing, Neither right nor left has the big dip at 48hz, but combined, they cancel.
Phase:
Left, right and both. 180 degree difference in phase at 48Hz.
Group Delay:
Left, right, and both
I have no idea how two positive group delays combine to create a big negative.
Step Response
Left, right and both
I don't notice any harm.
Hopefully the idea that there is, per se, such as thing as too much EQ has diminished.
The fave speakers of this forum (and others) like Kii, D&D, Neumann, etc only get their perfect responses by aggressive full range EQ.
The supposed danger is when you apply that approach to room EQ.
But I'm not sure that is the case these days. For a start, the supposed Schroeder FQ in any room I've had speakers in does not match where the response goes wonky. These days I do a measurement, see where the wiggles start and go from there.
Fortunately, the smart EQ systems do fine-grained stuff down low and sweepy EQ things above.
I think the new variable smoothing on REW is an excellent starting point.
The fave speakers of this forum (and others) like Kii, D&D, Neumann, etc only get their perfect responses by aggressive full range EQ.
The supposed danger is when you apply that approach to room EQ.
But I'm not sure that is the case these days. For a start, the supposed Schroeder FQ in any room I've had speakers in does not match where the response goes wonky. These days I do a measurement, see where the wiggles start and go from there.
Fortunately, the smart EQ systems do fine-grained stuff down low and sweepy EQ things above.
I think the new variable smoothing on REW is an excellent starting point.
I would separate the 2 ch vs Mch issue.
Spectral harmonizing by Eqing the whole frequency band might not be that big a deal with two speaker. They are mostly placed in similar sounding places and then there is no soundstage reconstruction from 2 speakers anyway.
For Multichannel the issue is different and I guess a lot of the new EQ interest comes from there. Not many people have 5+ identical speakers. And not many people have a perfect shoebox with a chair in the middle (and a transparent screen for video). And with 3 speakers in the front one might start illuminating the sound stage a bit if all is balanced well.
So my take is, that for Mch full frequency EQ & time alignment is great, for 2 ch it depends.
Spectral harmonizing by Eqing the whole frequency band might not be that big a deal with two speaker. They are mostly placed in similar sounding places and then there is no soundstage reconstruction from 2 speakers anyway.
For Multichannel the issue is different and I guess a lot of the new EQ interest comes from there. Not many people have 5+ identical speakers. And not many people have a perfect shoebox with a chair in the middle (and a transparent screen for video). And with 3 speakers in the front one might start illuminating the sound stage a bit if all is balanced well.
So my take is, that for Mch full frequency EQ & time alignment is great, for 2 ch it depends.
ernestcarl
Major Contributor
Thanks for sharing. It looks like at least one of the deep notches correspond with the negative delay. Everyone's listening space is incredibly different. It does seems to me that perfectly symmetrical rooms are harder to come by nowadays.
ernestcarl
Major Contributor
Like everything else, I do believe that there is such a thing as too much EQ. Sounds rather common sense to me. It's obviously not a prescription for one not to do EQ. But there's a very good reason why sound reinforcment guys don't use automated EQs willy nilly, but rather do it manually, and with much careful consideration to other variables.
I'm not sure I understand what you mean with your second statement. I have no idea about the Kii and D&D, but the KH120 already measures well minus the room -- and with zero EQ in place. I've already posted my manual PEQs for the KH120s here (for my listening space) and none of it looks overly aggressive/full-range to me.
ernestcarl
Major Contributor
I'm not sure if you've seen the post I did about the EQ I did for my LSR305 rears but it's very, very minimal. Only two PEQs and one -2dB treble shelving already built-in the monitors. If you're already starting with very linear monitors to begin with, you shouldn't need as much aggressive EQ correction.
I think the problem isn't too much EQ, but rather using EQ based on a room curve and not anechoic measurements. What Dr. Toole has said is that sometimes you can EQ something that looks like a problem in the steady state room curve but you could actually be doing more harm than good. I also have LS50s so my example will be relevant to the OP. Like many, I thought they could be a bit bright and fatiguing after longer listening sessions so I checked various measurements to find out why. In my case, my in room measurements are pretty close to what the anechoic measurements say need to be corrected, the biggest thing to correct is the 2k resonance but I took a bit around 750Hz and 5k as well. The key is making sure the EQ cuts you make isn't going to fix the on-axis sound while messing up the off-axis or vice versa. I actually found a "Spinorama" style measurement of the LS50 in a study which makes this analysis much easier. You can see by cutting at 2k, it not only fixes the listening window but the same peak is evident in all of the curves, making it safe to filter out. 750Hz is trickier because you can see the bump in the ER and SP curves is barely seen, in this case I only take out about 1.5 db and also listen about 15 deg off-axis which mostly takes care of it. I've also attached my in-room measurements that show the same peaks as the Spin and the before and after EQ measurement.
ernestcarl
Major Contributor
If you've eliminated a lot of the reflection points with treatment, you could probably EQ more of the higher frequencies without too much averse overall effect.
I'm pretty sure the KH80, like the Kii and D&D and the LS50W, uses a considerable amount of DSP correction to get the flat response it exhibits. This isn't room correction, it is speaker correction. I don't use full range room correction, but I do use full range speaker correction.
Similar threads
