EQ'd speakers are still too shouty

[kemmler3D]

Nobody's asked yet but do you have a THD measurement?

@Duke and @terryforsythe have also posted some plausible explanations for what you're hearing.

Diffraction (close up to the horn) might be overemphasizing some high frequencies, and if you look at the compression chart from Erin's review, you're getting a little boost at 2Khz every time there's an SPL peak, which would tend to be "shouty" as well.

And that 4dB peak around 1500hz could be contributing. Hard to EQ that out, but maybe some room treatments would help.

If you want speakers with a similar design that are more suitable for nearfield you might look at selling these and getting some JBL 305 or 308p - I usually see them pretty cheap secondhand on facebook marketplace. Generally people seem to like them, haven't heard anyone call them shouty.

 

[AnalogSteph]

If you want speakers with a similar design that are more suitable for nearfield you might look at selling these and getting some JBL 305 or 308p - I usually see them pretty cheap secondhand on facebook marketplace. Generally people seem to like them, haven't heard anyone call them shouty.

Mind you, plenty have called them noisy though... But yeah, there may be better alternatives for the desk. Kali LP6v2s are about the same price as a set of these Klipschs.

 

[klipschsoundsystem]

Mind you, plenty have called them noisy though... But yeah, there may be better alternatives for the desk. Kali LP6v2s are about the same price as a set of these Klipschs.

yeah i think im gonna sell them and get the lp6v2. i would have to sell the amp as well

 

[Rick63]

Is a username change in store for the OP?

 

[Mr. Widget]

FWIW (which, admittedly, ain't much):
In my four-plus decades of experience (albeit with occasional/rare exception): Klipsch = shouty.

I love Klipsch.
I love the legend, the thinking behind the designs, the aesthetics of their Arkansas built speakers, I respect PWK and Roy Delgado, but gosh darnit, like you, for the past 40+ years every Klipsch speaker I have owned I have sold off. (Heresy, a couple of versions Cornwall, La Scala, Klipschorn, and DIY Belle Klipsch) I find them all to sound too shouty for my taste.

 

[beeface]

I'm a layperson, so can someone let me know if this makes sense? if I said something with cupped hands around my mouth, then said the same the same thing with hands cupped around my mouth but in a different tone, it'd still sound like I had hands cupped around my mouth. right?

is this an appropriate analogy for horns + EQ?

 

[terryforsythe]

yeah i think im gonna sell them and get the lp6v2. i would have to sell the amp as well

Since you already have them, you might as well try equalizing them without the Harman curve to see if it at least helps. But, there probably are better speakers for the application.

 

[klipschsoundsystem]

Since you already have them, you might as well try equalizing them without the Harman curve to see if it at least helps. But, there probably are better speakers for the application.

my target (gray dotted line) is less bright than harman. i used to use -1db/oct tilt, it didnt sound that much better

 

[mhardy6647]

I love Klipsch.
I love the legend, the thinking behind the designs, the aesthetics of their Arkansas built speakers, I respect PWK and Roy Delgado, but gosh darnit, like you, for the past 40+ years every Klipsch speaker I have owned I have sold off. (Heresy, a couple of versions Cornwall, La Scala, Klipschorn, and DIY Belle Klipsch) I find them all to sound too shouty for my taste.

I concur, FWIW.
I like the notions that influenced the Klipsch loudspeaker designs, and Col. P. W. Klipsch himself was an utterly unique presence in hifi*, with a good bad attitude vis-a-vis the audio status quo and the "Establishment".

Klipschisms — Klipsch Museum Of Audio History

www.klipschmuseum.org

Spoiler

Klipsch owneship led me to discove good horns and drivers, so there is that.
PS I still maintain that the ultimate take on what Klipsch tried to do with the Heresy was Electrovoice's Esquire.



______________
* no mean feat, given the proliferation of 'characters' in the business, past and, yes, even present.

 

[Hoosierdaddy]

I'm a layperson, so can someone let me know if this makes sense? if I said something with cupped hands around my mouth, then said the same the same thing with hands cupped around my mouth but in a different tone, it'd still sound like I had hands cupped around my mouth. right?

is this an appropriate analogy for horns + EQ?

Yes....roughly speaking. But with some caveats.

Much depends on the actual design of the horn and it's implementation of course. The so-called 'cupped hands' coloration does seem to be more pronounced/noticeable in higher frequencies. Say upper midrange and higher. Also, the horn profile itself can strongly influence the 'honky' sound. As well as the material used. For example, generally speaking metal horns will have certain resonances which must be properly damped to avoid ringing.

But here we must apply a modicum of common sense. Horn loaded drivers have been a staple of the pro sound industry for many many years now. As I stated in my first post, horns are really designed to throw sound. To project it out across a room. There has been a good bit of research which shows that the further you get from a horn loaded driver, the more those colorations seem to disappear and the sound becomes more coherent. That's why I said they are not really designed for near field listening.

TO THE OP: Here's a test. Listen to your Klipsch from your usual listening position. Now walk across the room and listen - on axis of course. I'd be willing to bet there is a whole world of difference in how they sound. And that is my point.

 

[Duke]

I'm a layperson, so can someone let me know if this makes sense? if I said something with cupped hands around my mouth, then said the same the same thing with hands cupped around my mouth but in a different tone, it'd still sound like I had hands cupped around my mouth. right?

is this an appropriate analogy for horns + EQ?

Imo it's not that simple, and good horn design has advanced well beyond "cupped hands".

"Cupped hands" coloration is often a resonance caused by the reflection at the horn's mouth. The reflection at the mouth propagates partially outwards (away from the speaker) and partially back down the horn (towards the throat and compression driver). The back-down-the-horn portion of the reflection energy ping-pongs (resonates) back-and-forth between the mouth and the compression driver as it decays. Because this resonance arrives at the ears later than the original signal it is not "masked" by the louder original signal, so it tends to be more audible and objectionable than a look at the frequency response curve would predict. (I've simplified a bit for the sake of brevity.)

Let's see what happens when we short-circuit this reflection: Cup your hands around your mouth and say "aaaah". Keep saying "aaaah" while you move your hands forward away from your mouth by about half an inch. The "cupped hands" sound goes almost completely away, almost as much as if you remove your hands from your mouth entirely. What you did was, allow an escape path for that back-down-the-horn reflection so it couldn't build up into a significant resonance. (Again I've simplified a bit for the sake of brevity.)

Obviously what I described above won't work for a real horn, so imo the solution is to have a very gentle round-over at the mouth which minimizes the reflection back towards the throat.

"Cupped hands" can also arise from a hump in the frequency response which should have been fixed in the crossover. But crossover design and/or EQ cannot fix the resonance described above; that has to be addressed in the design of the horn itself.

 

[Tim Link]

My take looking at the response plot is that there's a significant recess in energy between about 80 Hz and 300 Hz if that dashed line is supposed to be the target. The peaks are above it but the dips are plentiful and they extend well below it. I'd try a plus 3 dB boost centered at about 150 Hz with a 2 octave width.

 

[Tim Link]

my target (gray dotted line) is less bright than harman. i used to use -1db/oct tilt, it didnt sound that much better
View attachment 408094
View attachment 408095

A way of looking at this target compared to the flat tilt is that you have a peak at around 1.5 kHz with dips at 200 Hz and 5 kHz. That looks like a potentially shouty response to me. But why didn't your other speakers come across that way? It'd be interesting to see the actual response you were getting on those compared to the target.

Looking at the MR4's spinorama I can see that it loses directionality up above 3 kHz which would counter the dip in your target response. The RPM 600 M2 don't do that but instead have a fairly narrow band where they lose directivity at 1.5 kHz.

I have some experience with EQing horns and also adjusting crossover settings and listening to the results. I have speakers that I can make very shouty or quite the opposite depending on where I set the crossover and how I adjust the EQ. When I hear what I think of as shouty speakers they have a lot of energy centered around 1.5 kHz.
I don't know how your EQ works but with mine I can slide a parametric EQ around while I'm listening and dial in on an effect I'm hearing pretty quickly. I'd start out with trying a 2 octave width parameter at minus 3 dB at 1.5 kHz. I'd be amazed if that didn't take some shout out of it. It might not sound good right there, but if you slide it around a little and adjust the strength you might find some thing that really works for you.

 

[Tim Link]

Let's see what happens when we short-circuit this reflection: Cup your hands around your mouth and say "aaaah". Keep saying "aaaah" while you move your hands forward away from your mouth by about half an inch. The "cupped hands" sound goes almost completely away, almost as much as if you remove your hands from your mouth entirely. What you did was, allow an escape path for that back-down-the-horn reflection so it couldn't build up into a significant resonance. (Again I've simplified a bit for the sake of brevity.)

Obviously what I described above won't work for a real horn, so imo the solution is to have a very gentle round-over at the mouth which minimizes the reflection back towards the throat.

I've been tinkering with my speakers, thinking about how to short circuit that mouth reflection on my exponential tweeter horn that doesn't have a nice roundover. I did it electronically with a delayed signal using REW's alignment tool. It's quite nice to push this little diffractive mouth horn all the way down to it's cutoff and not hear that resonance. I'm not the first to do this, but this is the first time I got it to work really well. REW's alignment tool makes it so much easier to figure out the low pass filter, delay and volume for the canceling signal.

I think it might be possible to clean it up more with multiple signals but this is already making an impressive difference.
The delayed signal has a low pass filter to keep it from disturbing frequencies where it's not needed.
It's attenuated 6 dB and delayed 0.095 ms to create the green result.
This really doesn't make the speaker more or less shouty. It lets it stay smooth and silky sounding even if run all the way down to it's rolloff at 1 kHz.

Spectrograms normalized to show reduction in resonance.

 

[Keith_W]

Re: "shouty" sounding horns. Earlier this year I read a paper on Gunness Focusing - paper attached.

In short, when sound travels from the throat to the mouth, it encounters a discontinuity. The sound reflects from the mouth, back into the throat with a slight delay determined by the speed of sound and the path length, and interacts with the direct sound producing phase cancellation effects. This is a so-called "two port" system, where there is one input port and one output port. Gunness lists some pre-requisites for a two-port system, and points out that it can be fixed with DSP.

I have tried measuring this discontinuity in my horn speakers. You will need a special test signal. You can not use a conventional swept sine wave because it is too long and does not have constant group delay. You need a sharp, short pulse - a Dirac delta function.

Then you examine your impulse response to look for a reflection. So I found what I thought was a reflection, occurring at 0.229ms after the main impulse. Some math tells me that 0.229ms = 78mm, which seemed rather early to me. I don't think it's an internal cabinet reflection, since the cabinet is pretty large. I wasn't sure what I was looking at, but I decided to proceed anyway. I designed a filter to cancel that response in the same way that I design a DSP filter for a VBA (thread).

Here is the correction filter I designed, in green.

And here is the simulated result in black, compared to the original signal. It looks a little better.

Unfortunately, I did not take a measured result. What did it sound like? It was really weird. My speakers are normally quite neutral, but this introduced a kind of honking, nasal quality to the sound. I hated it so much that I gave up on the experiment in disgust and did not say anything about it on ASR until now.

There are a few potential reasons why this failed. The main one is, don't try to screw around with speakers that sound fine in the first place to correct a "problem" I see in the measurement, whose cause I was unsure about. The worst that happened to me was a few days of wasted time. I don't see that as a problem, thing is - NOBODY knows what algorithm David Gunness used. The idea is patented, but the actual procedure is secret.

 

[AnalogSteph]

Here is the correction filter I designed, in green.

That seems to be doing a lot more than just taking out the reflection? You didn't cross-check it in frequency domain (amplitude and phase)?

If the reflection is real, it should have been visible in the form of periodic ripple in frequency domain, with the first dip at about 2.18 kHz and first peak at 4.36 kHz if I'm not mistaken.

 

[Tim Link]

Whatever that bump is, is there any sign that it's repeating further out on the impulse like mine was out to 4 milliseconds?

 

[Keith_W]

That seems to be doing a lot more than just taking out the reflection? You didn't cross-check it in frequency domain (amplitude and phase)?

Yep, I did. Not much to see there.

 

[Keith_W]

Whatever that bump is, is there any sign that it's repeating further out on the impulse like mine was out to 4 milliseconds?

No. Which is what made me wonder what the heck it was.

 

[Tim Link]

I've been tinkering with my speakers, thinking about how to short circuit that mouth reflection on my exponential tweeter horn that doesn't have a nice roundover. I did it electronically with a delayed signal using REW's alignment tool. It's quite nice to push this little diffractive mouth horn all the way down to it's cutoff and not hear that resonance. I'm not the first to do this, but this is the first time I got it to work really well. REW's alignment tool makes it so much easier to figure out the low pass filter, delay and volume for the canceling signal.

I think it might be possible to clean it up more with multiple signals but this is already making an impressive difference.
The delayed signal has a low pass filter to keep it from disturbing frequencies where it's not needed.
It's attenuated 6 dB and delayed 0.095 ms to create the green result.
This really doesn't make the speaker more or less shouty. It lets it stay smooth and silky sounding even if run all the way down to it's rolloff at 1 kHz.
View attachment 500443
Spectrograms normalized to show reduction in resonance.
View attachment 500442View attachment 500441

Guess what!
The cancel signal worked, but not for the reason I thought.
I decided to test this idea to the extreme by using a 1 foot long 1" diameter PVC pipe as a horn. It resonanted badly, and I fixed the resonance with the cancel signal. Once EQ'd it actually sounded pretty nice, with fantastic dispersion. Unfortunately there's little to no efficiency gain in such a device. The exit shock causes a huge loss of energy at lower frequencies, so there's no point in using a straight tube as a horn except to play around with canceling its resonances. I'll have to admit I was hoping there'd be at least a little gain from the length of the tube. Horns are about the mouth size! The only reason they need length is to smoothly transition to a bigger mouth.

So yes, the idea works on PVC pipe, but it turns out my tweeter horn is not nearly so bad, and doesn't really have any audible or even visible resonances on the spectrogram. So what caused the resonance so clearly visible in my earlier measurements? Turns out it's the sound from the tweeter bouncing around inside the larger midrange horn that it is coaxially mounted in. I figured this out because the resonance disappears when I measure off axis. My cancel signal successfully removes it on-axis, and it actually sounds better. But this situation cannot be fixed for all listening angles with a cancel signal because it's not a 2 port problem.

The real solution for my situation is some absorption on the inside of the mouth of the midrange horn. I'll see if I can get that to work and sound good.