Klipsch Roy Delgado explaining why Klipsch measures so bad.

[gnarly]

I dunno if Dirac is doing anything analogous to what's in that paper, but "correcting" the impulse response is indeed one of the things Dirac attempts to do.

Yep. Most all FIR correction filter usage appears to come from some form on impulse inversion. Certainly so for automatic type corrections.

I think Fulcrum Acoustic's 'Temporal EQ' as they call it, is one of the simplest impulse inversions out there. It's been around for quite a while...like 2005/6.
A short impulse inversion, of a horn with a compression driver. Corrects internal horn reflections.
Something that has high value for all horn/CD combos, home audio and PA, imo.

Dirac I think, is quite a bit more complicated. We know they tout mixed phase FIR utilization. I've been trying to determine just how, by making transfer functions of the electrical filters Dirac generates. It appears that in addition to whatever frequency domain corrections are needed (the usual min phase smoothing), Dirac attempts to match the lower frequency phase response of one side to the other. On mine, Left side gets left pretty much alone, and Right gets its phase rotations aligned to more closely match Left.
So it does not appear to me that Dirac is trying to reduce phase rotation; it appears it's trying to match existing rotations side to side.
That said, I could be off about it not trying to reduce phase rotation, because my speakers much have less phase rotation than is common. I'll need to give Dirac a problem child to really find out Lol

Anyway, just saying I agree with your take about FIR corrections being rooted in impulse inversion.

 

[Cbdb2]

The facts are that Heritage Klipsch is quite popular and enjoyed by many thousands of happy owners, even more so today than in the 75+ year life of the company.

The fact is popularity and performance are not related, especially in audiophile land.

 

[NTK]

Where's the flatness??

Or low distortions. Heresy IV vs KEF R5.
Source: https://www.erinsaudiocorner.com/loudspeakers/

 

[aschen]

Heritage series can play very loud especially with limited power, and are good looking MCM aesthetic that is so popular these days

Seems like two very salable features when your not bound by "blind and level matched"

 

[Chris A]

...Or low distortions. Heresy IV vs KEF R5...

There is a difference in price of 1.66:1. You guess which one is more expensive. I assume you remember this:

https://www.audiosciencereview.com/...sch-measures-so-bad.62444/page-3#post-2290884

Additionally, I have no interest in Heresies or defending the harmonic distortion levels of them. However, if one were to show modulation distortion levels (particularly AM distortion), that might be more interesting if evaluated above 100 dB/1m, which is the starting point to see a real difference.

The fact is popularity and performance are not related, especially in audiophile land.

I think you took that quote out of context. Look again at the statement I was addressing. It is relevant.

Just as a request: if you're going to lift statements out of context, I ask a bit more "due diligence" than that. It appears like you're in some sort of rhetorically fueled debate.

I hope that's not what this forum is about.

Chris

 

[Chris A]

I think Fulcrum Acoustic's 'Temporal EQ' as they call it, is one of the simplest impulse inversions out there. It's been around for quite a while...like 2005/6.
A short impulse inversion, of a horn with a compression driver. Corrects internal horn reflections.
Something that has high value for all horn/CD combos, home audio and PA, imo.

Except that, when I asked Greg Berchin about that JAES report by Gunness, he immediately made it clear that this wasn't something that he was conversant with, and that it was well out of his "bag of tricks".

In other words, I think you're misinterpreting what Dave Gunness is actually doing. He formed a company (Fulcrum Acoustics) around the patent that I posted the link to, and the last time I looked, they were doing well.

When I worked for a large geophysical contractor in the late-70s/early-80s, I worked with a PhD (M.E.) locally and a PhD mathematician in the main office in Houston to do the same thing, but in this case it was to remove excessive harmonics from the measurement signal sweep of large Vibroseis machines (that I was a design engineer on) that all experienced "pressure feedback" anomalies through the main servovalve that amplified the incoming electronic signal to produce its "vibrator" output (32-77 thousand pound peak force). The nonlinearities of the system and servovalve in particular were modeled and checked against real data, then "inverted" (a nonlinear operation that was found to be invertible mathematically under certain conditions). They found the right preconditioning waveform for the input waveform to significantly reduce harmonic distortion at the lowest sweep frequencies, but it had a discontinuity in its shape that was difficult at that time to handle using then-current field electronics (TI and Pelton Instruments). Also about then (1981) the bottom fell out of the exploration market worldwide, so that work was shelved. I don't know if it was submitted for patent at that time, but it was certainly patentable.

This point is, this isn't your typical nonlinear system inversion problem. And the equipment that Gunness uses doesn't run off-the-shelf DSP audio software, either.

Chris

 

[PristineSound]

It appears (yet again) like I'm dealing with a heavily spring-loaded opinion here that doesn't change, no matter what information to the contrary is presented. While I am more than willing to talk about specific details of pros/cons of Heritage Klipsch models, making these sort of sweeping generalizations I believe does no one any service, and in fact, looks quite biased. I don't believe I'm the only one here that has formed that opinion over the less than one week duration of this thread.

It's okay to have decision bias, but for me, it's not okay to keep pressing it in such an uncalibrated fashion on an open forum such like this--which is dedicated to "the science of home hi-fi audio components". Your above quoted statement can be your opinion, but it certainly isn't fact. The facts are that Heritage Klipsch is quite popular and enjoyed by many thousands of happy owners, even more so today than in the 75+ year life of the company.

Aside from calling you out on blaming conspiracy theory or discrediting the multiple various members, including myself, who are poking holes in a story and not buying it, nothing else needs to be said (edit: "to you." that is. I still like to discuss with others, especially on MEH). This thread is available here for anyone to read, it includes your "specific details". It's transparent and available to any reader.

 

[PristineSound]

Or low distortions. Heresy IV vs KEF R5.

$600 SVS Prime, just some random speakers I so happened to click on on Erin's website.

 

[gnarly]

Except that, when I asked Greg Berchin about that JAES report by Gunness, he immediately made it clear that this wasn't something that he was conversant with, and that it was well out of his "bag of tricks".

In other words, I think you're misinterpreting what Dave Gunness is actually doing. He formed a company (Fulcrum Acoustics) around the patent that I posted the link to, and the last time I looked, they were doing well.

Hi, I can't see how, what Greg knows or doesn't know, has anything to do here. I've had many fine discussions with him over the years on various forums, and have great respect for his knowledge of filter and crossover implementation, particularly IIR. Use of FIR, specifically for time domain corrections, is not something I've gathered he is as familiar with.

Dave G certainly didn't build a biz mainly around that patent. He had been with EAW before leaving to form Fulcrum Acoustics. At EAW, he was already well known in the prosound industry for "Gunness focussing" ...presets in the EAW UX series processors that included undisclosed "tricks".
It helps to remember that he was one of the first pioneers into using FIR in prosound processors. This is old stuff and must say I think you are over interpreting what he is actually doing.
Dave has built a successful business because he's a damn fine speaker builder and tuner...starting with the acoustic design! Plus FA apparently has a great bunch of people.

I had the good fortune to briefly meet him years back and jokingly converse with him that I was working on cracking 'Temporal EQ', while being serious at the same time.
I was relatively new to FIR then, particularly so for corrections where the emphasis was on the time domain.. After a number of experiments, I realized I was making it much harder than it is ...
fwiw

 

[kyuu]

$600 SVS Prime, just some random speakers I so happened to click on on Erin's website.
View attachment 448109

View attachment 448110

Or my speakers, MSRP $495 (and can be obtained at a significant discount from a dealer):

 

[gnarly]

This point is, this isn't your typical nonlinear system inversion problem. And the equipment that Gunness uses doesn't run off-the-shelf DSP audio software, either.

Uh, it is typical gear. FA imbeds the 'temporal EQ' settings into a variety of prosound processors that have FIR, so folks can use their speakers with whatever processing infrastructure they already own. https://www.fulcrum-acoustic.com/support/processor-configurations/
A person can see for themselves what the processing files do, if they have any of the listed processors.

 

[Chris A]

The issue is not the final implementation of the filters in either IIR or FIR format, it's the underlying model of how you get to the final filters.

Gunness (and the two gentlemen I discussed in the geophysical example) show pretty extensive non-linear dynamics models that they first calibrate via measurements, and then load in the inversion filters after those detailed models are complete and calibrated to the exact instance/driver. That's what different. Depth of knowledge of the nonlinearities and deep modeling.

Using only surface-level measurements of the nonlinearities and then just generating FIR empirically, without any understanding of what's actually causing and then modeling those nonlinearities (like the multiple slit phase plug example, or the effect of reflected impedance bounces in a coaxial driver horn due to a sudden change in expansion back through the compression driver)--these are the details that make or break the degree of success in linearization of very non-linear phenomena in an acoustic driver compensation system.

That's why I think Greg backed away from (as I did, since the next step in modeling is a very deep dive into the pool). It's the depth of understanding the source of the nonlinearities and the complexity of modeling them and calibrating those models that sets that sort of work apart--not just using something like RePhase to produce FIR filters without any true understanding what is actually occurring.

YMMV.

Chris

 

[gnarly]

The issue is not the final implementation of the filters in either IIR or FIR format, it's the underlying model of how you get to the final filters.

Gunness (and the two gentlemen I discussed in the geophysical example) show pretty extensive non-linear dynamics models that they first calibrate via measurements, and then load in the inversion filters after those detailed models are complete and calibrated to the exact instance/driver. That's what different. Depth of knowledge of the nonlinearities and deep modeling.

Using only surface-level measurements of the nonlinearities and then just generating FIR empirically, without any understanding of what's actually causing and then modeling those nonlinearities (like the multiple slit phase plug example, or the effect of reflected impedance bounces in a coaxial driver horn due to a sudden change in expansion back through the compression driver)--these are the details that make or break the degree of success in linearization of very non-linear phenomena in an acoustic driver compensation system.

That's why I think Greg backed away from (as I did, since the next step in modeling is a very deep dive into the pool). It's the depth of understanding the source of the nonlinearities and the complexity of modeling them and calibrating those models that sets that sort of work apart--not just using something like RePhase to produce FIR filters without any true understanding what is actually occurring.

YMMV.

Chris

I think you are way overthinking what's going on.
And it is simply a matter of measurements that can be acted on, and THAT is the name of the game.
Far more so than modeling, or mental hypothesizing. Accepting, using, empirical measurement results is not such a deep dive.
Just spend some time measuring horn/CD combo impulses, and analyzing the time window interval horn reflection are possible within.
Mount a mic in the horn close to the CD to catch reflections, or even better get a coax CD and use one section as driver and the other as microphone.
Both techniques will give you a handle to see it's much more akin to simple impulse inversion .

 

[Chris A]

Mark, I can see that everything is still very simple...

...that is, until it's not.

Chris

 

[gnarly]

Mark, I can see that everything is still very simple...

...that is, until it's not.

Chris

Chris, for me...until I see it simply...i've learned i don't yet fully understand it......

 

[Travis]

Yes.

"Klipsch jest jednak aktywnym partnerem EISA, a więc stowarzyszenia ekspertów, a nie influencerów, i bez oporów dostarcza nam do testów wszelkie produkty"

Translation:
However, Klipsch is an active partner of EISA, i.e. an association of experts, not influencers, and they provide us with all their products for testing without hesitation.

Unfortunately that statement is complete and utter bullshit. Klipsch never sent them a set of Jubilees to test. Klipsch isn’t a partner in EISA, but John Darko is if that tells you anything.

I remember when that was first posted here and the consensus was there were many unanswered questions on how they tested, equipment, used and an assortment of other things was entirely lacking.

@Bjorn can probably look at that and explain why the curves they claim don’t match with the extensive testing he has done on the 402 horn.

 

[NTK]

Yes, in theory. Infinitely big volume means infinite area of reflecting walls hence indefinitely sustaining indirect sound even if there is wall absorption. Note that Sabine's formula is neglecting the adiabatic properties of air, i.e. increasing absorption towards smaller wavelengths, so in reality this will never happen. And of course in reality there is an initial time delay between direct and indirect sound at play.

If you want to have a glimpse of an idea how it sounds, take a clicker frog as used by recording engineers to a large empty water cistern (or a cathedral). It's fun.

The term "adiabatic" is used to describe thermodynamic processes that have no heat transfer between the thermodynamic system and the outside. It is a necessary condition for the reversible thermodynamic process — a theoretical concept (and the other being isentropic, i.e. constant entropy). I have never seen it being used to describe "properties" of a fluid.

Anyhow, the primary cause for the attenuation of sound waves in air is due to the "volume viscosity" (as opposed to the dynamic viscosity, and there is no simple direct relationship between the two), which is the result of the fluid medium not having sufficient time to reach thermodynamic equilibrium during the compression/expansion process. That also explains why the effect is much more pronounced at higher frequencies as the time scales of the compressions/expansions are much shorter. It doesn't, AFAICT, have anything to do with any "adiabatic properties" of air.

Here is the chapter 1 section the above screen clip referred to:

https://sjoerdr.win.tue.nl/papers/boek.pdf

 

[Travis]

Yes.

"Klipsch jest jednak aktywnym partnerem EISA, a więc stowarzyszenia ekspertów, a nie influencerów, i bez oporów dostarcza nam do testów wszelkie produkty"

Translation:
However, Klipsch is an active partner of EISA, i.e. an association of experts, not influencers, and they provide us with all their products for testing without hesitation.

This is also in the article, per my neighbor’s wife, an EE who was born and raised in Poland until she was 18 who reviewed and corrected an Apple Translation.

“The second "organizational" issue requiring presentation is the setting of the microphone. Due to the pulsed method of measuring characteristics in the mid-high-range range, the microphone cannot be more than 1.5 meters away; why? - we will not explain this time, because we have a lot to write about the Jubilee itself anyway.

When measuring two-way structures, we use a distance of 1 meter, usually sufficient for good integration of transmitter radiation, but here... although it is a two-way system, it seems to require a greater distance due to its dimensions.

The woofer section is combined with the midrange to the high section at a height of 110 cm, and using the classic method of deriving the main measurement axis from such a point of construction, it is more or less the same as the probable height of the ears of the seated listener.

However, it may be doubtful whether the microphone located at such a height at a distance of only 1.5 meters is not at too large an angle in relation to the main axis of the tube itself, which runs at a height of as much as 140 cm, which could cause a drop in pressure, especially at the edge of the band.

Even keeping the height of 110 cm, but moving away from the column, we reduce this angle and potential loss. However, we checked that it is (the difference between the characteristics of the mid-high-range measured at a height of 110 cm and 140 cm) small, within 1 dB (so the dispersion of the tube is very good).

However, still measuring the sound source with such a large surface area (even exactly on its axis of symmetry) from a moderate distance (slightly larger than its diameter) makes mistakes. Perhaps this is the reason why the measured characteristic has clearly exposed high frequencies, and at a greater distance this effect would be less thanks to (relative!) increase in medium tone pressure. We can't help it anymore, because increasing the distance (microphone from the column) adapted to the measurement method would cause even greater errors.”

(Emphasis added). Again, John Darko is one of their “partners.”

Katrina’s comment (BS from Rice, MS from UT in Electrical Engineering) “from beginning to end in this article they say the testing has errors, they know it has errors. The only thing they felt sure of in the measurements was that it was -6 dB at 16Hz. I would be embarrassed to extract a measurement graph from that article and post it anywhere where an engineering or scientific discussion was taking place.”

 

[PristineSound]

This is also in the article, per my neighbor’s wife, an EE who was born and raised in Poland until she was 18 who reviewed and corrected an Apple Translation.

“The second "organizational" issue requiring presentation is the setting of the microphone. Due to the pulsed method of measuring characteristics in the mid-high-range range, the microphone cannot be more than 1.5 meters away; why? - we will not explain this time, because we have a lot to write about the Jubilee itself anyway.

When measuring two-way structures, we use a distance of 1 meter, usually sufficient for good integration of transmitter radiation, but here... although it is a two-way system, it seems to require a greater distance due to its dimensions.

The woofer section is combined with the midrange to the high section at a height of 110 cm, and using the classic method of deriving the main measurement axis from such a point of construction, it is more or less the same as the probable height of the ears of the seated listener.

However, it may be doubtful whether the microphone located at such a height at a distance of only 1.5 meters is not at too large an angle in relation to the main axis of the tube itself, which runs at a height of as much as 140 cm, which could cause a drop in pressure, especially at the edge of the band.

Even keeping the height of 110 cm, but moving away from the column, we reduce this angle and potential loss. However, we checked that it is (the difference between the characteristics of the mid-high-range measured at a height of 110 cm and 140 cm) small, within 1 dB (so the dispersion of the tube is very good).

However, still measuring the sound source with such a large surface area (even exactly on its axis of symmetry) from a moderate distance (slightly larger than its diameter) makes mistakes. Perhaps this is the reason why the measured characteristic has clearly exposed high frequencies, and at a greater distance this effect would be less thanks to (relative!) increase in medium tone pressure. We can't help it anymore, because increasing the distance (microphone from the column) adapted to the measurement method would cause even greater errors.”

(Emphasis added). Again, John Darko is one of their “partners.”

Katrina’s comment (BS from Rice, MS from UT in Electrical Engineering) “from beginning to end in this article they say the testing has errors, they know it has errors. The only thing they felt sure of in the measurements was that it was -6 dB at 16Hz. I would be embarrassed to extract a measurement graph from that article and post it anywhere where an engineering or scientific discussion was taking place.”

Fair. Can someone share a third party measurement that has a high level of certainty on it's accuracy on any one of the Klipsch heritage line speakers?

 

[middlemarch]

As someone who built a set of Speakerlab Model Ks back in 1974 and then schlepped those monstrous piles of lumber around for almost twenty years, and having followed most of this thread, I for one don't regret leaving the whole Klipsch (or clone) scene in the rear view mirror. Good riddance. I still have the mid range horns and tweeters. Free to a good home. Won't ship.