Horn Speakers - Is it me or.......

[garyrc]

I love the sound of my Klipschorns (AK5, c. 2002 with Klipsch stock upgrade), in nearly airtight corners, with Audyssey Flat room correction plus tweeking with tone controls, a 4,500 cu.ft room with room treatment including painstakingly adjusted diffusion View attachment 389088 and absorption View attachment 389089, firm walls [(exterior panels, then 3/4" ply, 2X6 studs, glass wool, second layer 3/4 ply (staggered seams), and 5/8" Sheetrock, covered with rough plaster (except where it would hold K-horns out from the wall).

Some of my earlier response seen above (re: my Klipschorns w/ Audyssey room correction and room treatment) didn't come through for some reason. Here is the gist of the rest:

1. My K-horns have never sounded "cuppy."
2. My K-horns have never sounded "honky" to my ears, anyway. In fact, I noticed that the frequency range called "honk" on the one published chart I have is 450Hz to 1000Hz, a place where the K-horns more or less hug the line (+/- 2.2dB) on my own curve (w/Audyssey and REW) and Klipsch's factory curve (anechoic, except for a wood corner, since that is part of the design). On the Klipsch curve there is one very narrow dip, 1/10 of an octave wide, at 600Hz, supposedly inaudible at that width. So, I'll go with "no honk heard or seen."
3. The dynamics would be tough to beat. First, they provide crescendos that are pretty overwhelming, and peaks burst or plunge out mightily in a way that exemplifies Berlyne's cortical arousal increasing variables of surprise, complexity, and, of course, loudness, better than any speaker that I have heard other than the ancient ones I will show below. Secondly, in my room, a typical speaker (90 dB, 2.83v, 1m) would need about 128 watts to reach THX's 105 dB above 80Hz in a very large room. The K-horns would need only about 12 watts to provide the same 105 SPL, and if supplied with the 128 watts that produces the 105 dB with a typical speaker, the Klipschorn would throw 116 dB at us, which is about the SPL of the loudest instaneous peaks of a symphony orchestra. All this is given the likely "true" sensitivity of the K-horns of 101dB, 2.83v,1m -- Klipsch rates them at 105dB, adding the 4 dB to the anechoic measure to reflect the sensitivity in a "typical living room" and, I imagine, to avoid making it seem that the K-horn has lost sensitivity (it hasn't) since the earlier times when it was rated at 104 dB at 4 feet, in a typical living room, or about 54 dB EIA in the olden days.
4. If you are lucky enough to encounter Klipschorns (or the new Heritage Jubilee, which is said to be even better) in a store, know that in my wanderings only 43% of Klipschorn pairs have been set up correctly. The older ones need to be pushed fully into a corner, with an air seal, and the new ones need to be very close to a corner.
5. In the past, when high fidelity and stereo were young(er), there were some horn speakers that could really move air. I experienced them in 70mm, 6 channel sound movies. They could shake the concrete floor of the theater (the thunderstorm during the crucifixion in Ben-Hur, 1959 and 70mm re-release in 1968). There were no subwoofers, and the big horns did not have a wide frequency response (40 to 10K HZ, as high as 12K or higher with a few, and attenuated bass from just below 40 down to JBL's "usable to 32Hz"). But, man were they clean, tight, with great transients! IMO, they make some newer THX systems sound muddy!
   
   

 

[garyrc]

Klipschorn AK6 measured by Stereophile:

Klipsch Klipschorn AK6 loudspeaker Measurements

Sidebar 3: Measurements Because of the Klipschorn AK6's bulkeach weighs 220 lbI drove my test gear the 177 miles to Art's place and measured the speaker sitting on a furniture dolly in his driveway. I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the Klipschorn's...

www.stereophile.com

Stereophile's red curve should be ignored. The Klipschorn A6 requires positioning it NEAR a corner, and also needs the boundary gain of being on the floor, NOT on a raised furniture dolly in a driveway.

 

[Bjorn]

I've measured the Klipschorn in a corner. Let's just say it has serious room for improvement.

 

[rank]

With ABX, we can clearly see that people can love speakers that measure poorly, even at high volume where the distortion becomes rather high.

 

[Newman]

Please show me that.

 

[garyrc]

I've measured the Klipschorn in a corner. Let's just say it has serious room for improvement.

As does every speaker I've heard that I can afford. Of course, different listeners have different priorities. Clean, realistic dynamics loom large for me. Frequency response, not so much, especially with good room treatment and if the room and speakers are EQ'd to taste manually after using a room correction device, such as Audyssey, Dirac, Trinnov, etc.

I value:

• Those realistic dynamics, including instantaneous peaks (1/4 second, or less) of 100 dB to a conceivable 115 dB, even though the typical “loud” level is only about 90 dB.
• Low distortion, especially modulation distortion, even at very high SPL.
• Clean, fast transient response.
• High clarity and detail.

Lower priority (depending on the recording, in some cases).

• Good imaging and soundstaging
• The listener's musical gestalt, a la J. Gordon Holt. He posited that professional musicians want their speakers to trigger their musical gestalt, and reproduce the tonalities of musical instruments with low distortion, and have the ability to fill a room with a crescendo. I don't know how to measure this gestalt, but I think I've heard it often, especially while listening to much live classical music (my usual lunch break) and while playing in several orchestras.
• Qualities of the best vintage speakers that have been missing in several newer ones (from the mouth of Art Dudley) are: “Impact. Pluck. Snap. Body—especially body. And soul.”

I'm getting the impression that there are large random components in the frequency smoothness of recordings, depending on what the mixers actually hear. I'm sure most of you have seen these graphs of the frequency response of various dubbing theaters and control rooms.

What about speakers that are considered “flat?” What flat is has evolved. In 1967 the
AR3A was 7 dB down at 30 Hz, as opposed to -15 dB with the big JBL twin woofer horn (C55). I could see what the excitement was about, --- all that bass from that little AR box. In the 70s, High Fidelity magazine rated the AR LST “Laboratory Standard Transducer” at +/- 6 dB, “a good mark for any speaker.” Nowadays, “essentially flat” is more like +/- 2.75 dB. The YG, at more than $100,000, is rated by them at +/- 1 dB 20 to 20K Hz, but in a real room, when it was tested in a Stereophile reviewer’s listening environment it came in at what looks to me to be more like about +/- 4.5 dB (+4dB, -5 dB), 20Hz to 15K Hz, RED trace.

O.K., let’s turn to a horn speaker everyone knows isn’t flat, but has low modulation distortion, extraordinary dynamics that will throw my desk out of square, and is clean and detailed, with electronic room correction, using 8 microphone positions.
Rather than bring in a sub at 80Hz or so, since the Klipschorn is much cleaner than the sub down to about 50Hz, we crossover at 50Hz. We can therefore ignore the big bump at 45Hz, since the bump doesn't show up much through the sub (due to position?)

We have +3, – 2 dB (+/- 2.5 dB), 50 to 15K Hz, with the sub working below.

Same set up, but for very bass shy recordings:

Finally, what about speaker distortion? The Klipschorn was measured by the late, great Heyser as having 1.7% IM and 0.25% THD, both at 105 dB. The Fried Studio 4, I think, was measured by someone else at Audio, and had 10% IM, and 4% THD also at 105 dB. The others were measured by others with different, but good reps. Wish I still had the references. The AR 4 way AR 98RS had 2.7% IM and 3% THD, and the Platinum Studio 2 had 7% IM and 1.9% THD.

Let's go listen to some music!

 

[Newman]

As does every speaker I've heard that I can afford. Of course, different listeners have different priorities. Clean, realistic dynamics loom large for me. Frequency response, not so much, especially with good room treatment and if the room and speakers are EQ'd to taste manually after using a room correction device, such as Audyssey, Dirac, Trinnov, etc.

Hmmm. You may well be a rare outlier, but the best available audio science contraindicates the bit in bold above, and also contraindicates that equalisation will be subjectively effective on a speaker with the non-uniform off-axis behaviour of the K'Horn.

Your journey may have taken you down a sonic rabbit hole, and your subjective satisfaction with it may be largely confirmation bias. That's a much more likely explanation than you being an outlier who likes a poor FR and doesn't mind the result of subjectively-objectionable EQ. Because a huge number of audiophiles highly value clean realistic dynamics, you are not at all unusual in that regard, but tests have shown that low colouration, low resonances, and accurate FR are basically non-negotiable. There are much better ways to get that with more modern horn options.

cheers

 

[Postlan]

When comparing a rectangular horn (like Kinoshita and JBL) with a circular horn, what are the advantages and disadvantages of each?

 

[Holmz]

Hmmm. You may well be a rare outlier, but the best available audio science contraindicates the bit in bold above, and also contraindicates that equalisation will be subjectively effective on a speaker with the non-uniform off-axis behaviour of the K'Horn.

Your journey may have taken you down a sonic rabbit hole, and your subjective satisfaction with it may be largely confirmation bias. That's a much more likely explanation than you being an outlier who likes a poor FR and doesn't mind the result of subjectively-objectionable EQ. Because a huge number of audiophiles highly value clean realistic dynamics, you are not at all unusual in that regard, but tests have shown that low colouration, low resonances, and accurate FR are basically non-negotiable. There are much better ways to get that with horns today.

cheers

I thought that the @garyrc fellow said that they could be corrected?
Other things may not be as easily corrected digitally.

 

[Duke]

When comparing a rectangular horn (like Kinoshita and JBL) with a circular horn, what are the advantages and disadvantages of each?

Oboy. This is going to be a longer answer than you were hoping for, and it's still going to be oversimplified and incomplete. I'm not going to cover Ton Danley's Unity (or Synergy) horns, but I am going to list four general types of rectangular horns and two general types of round horns.

Most rectangular horns have a wider pattern in the horizontal plane than in the vertical plane, so they are generally better at covering the intended listening area than round horns.

1. Some old-school rectangular horns use vanes to get wider high-frequency coverage than they otherwise would have. The vanes are not as good at high frequency pattern-widening as a diffraction slot, but they don't introduce the coloration that a diffraction slot can. They tend to have good low-end loading so the crossover can be pretty low. The high frequency coverage tends to narrow in the vertical plane as we go up in frequency. Kinoshita, TAD, Augspurger.

2. Rectangular diffraction horns have a sharp-edged slot and/or a sharp-edged kink in the horn profile and/or sharp-edged lips at the mouth. These tend to have the most precise pattern control but arguably tend to have the most “horn sound" coloration, at least at high SPLs. They often have very good low-end loading because they can be quite deep. Not going to mention any by name 'cause I just dissed this kind of horn and theoretically I'm a competitor.

3. Rectangular “gentle diffraction” horns have a gentle slot or other gentle diffractive feature. They can combine good pattern control (though not as good as a sharp-edged diffraction horn) with low coloration. Good low-end loading. JBL M2, 4367, and 4349, Unison Research.

4. Rectangular “non-diffractive” horns. Low coloration but can beam in the top end, typically poor low-end loading, so they can't be crossed over too low unless they are pretty deep. Klipsch Jubilee, PBN M1!5, Volti Razz, Pi Speakers 4Pi.

Round horns tend to fall into one of two categories: Those that emphasize efficiency over pattern uniformity, and those that emphasize pattern uniformity over efficiency. They tend to be low-coloration devices.

1. Round non-constant-directivity horns have a profile somewhat reminiscent of a trumpet's flare, and include tractrix, exponential, hyberbolic, spherical, and Le Cleac'h horns. These are the highest efficiency and therefore most dynamic horns, as they concentrate the (scarce relative to the rest of the spectrum) high frequency energy on-axis. They also have the smallest sweet spot for the same reason. They tend to be quite deep and therefore have good low-end loading. This type typically needs no EQ aside from maybe some response smoothing. Avantgarde, Odeon, Edgarhorn (discontinued).

2. Round constant-directivity horns, whose profiles look more like a shallow cone with a mouth round-over, are the lowest efficiency horns. Their patterns are often not as uniform as the diffraction or gentle diffraction horns, and they probably have the worst low-end loading. They tend to beam somewhat at high frequencies. On the other hand they offer a good combination of low coloration and good pattern control over most of their range. This type typically needs aggressive EQ because its native response will be strongly downward-sloping. Dutch & Dutch, Gedlee (currently out of production but possibly coming back, presumably with improvements).

Horn design is a juggling of tradeoffs, macro and micro.

 

[kemmler3D]

large random components in the frequency smoothness of recordings, depending on what the mixers actually hear.

I'll just say that this graph shows a variation of maybe +/- 3, 5db at worst and the K-horns have multiple 10dB swings... And of course mixing / mastering engineers reference their mixes to other "known good" mixes, so they actively compensate for their listening setups that way. I don't think "the recordings themselves are just as flawed as the speakers" really washes.

Not saying the K-horns are bad per se, just saying you can't just hand-wave the FR.

 

[tmuikku]

Sorry Duke I'll expand your list a bit, having been following Marcel's thread in diyaudio.com: loading and pattern control can be adjusted separately to some extent, so one can have relatively wide pattern with more or less loading, and without diffraction. Narrow pattern devices typically come with loading and perhaps it cannot be taketh away. Diffraction slots are not needed for this, because the shape can be optimized to have same effects but without adverse effects of diffraction. At least it looks like so.

Rectangular and round shape make minor difference as such, same profile (curvature) on both makes about the same directivity and loading. While there is difference, basically it's not relevant amount and shape could be chosen by looks / utility. Shape of mouth could be almost anything sensible, not limited to round and rectangular shapes.

Both round and rectangular shaped device can be asymmetric and usually we see vertical dimension smaller than horizontal. This enables narrow pattern on vertical and wider pattern for horizontal as you say, but only for high frequencies down to roughly frequency whose wavelength is of the mouth short dimension. The shorter dimension loses pattern control higher up in frequency while the bigger dimension still has it, reversing the situation, now wider dimension has narrower pattern and shorter has wider. This is called pattern flip and is the trade-off with such devices. Perhaps it's an issue with highly asymmetric devices, like more than 2:1 ratio with the dimensions.

 

[rank]

Please show me that.

These listening sessions were conducted decades ago, with listeners who are mostly dead now. They are extremely common among sound worshipers who gather in hifi clubs, frequent hifi salons or the European triode festival.
Even if you have never been part of this kind of circles, you must have experienced this kind of experience at least a little bit, right?

 

[Newman]

These listening sessions were conducted decades ago, with listeners who are mostly dead now. They are extremely common among sound worshipers who gather in hifi clubs, frequent hifi salons or the European triode festival.
Even if you have never been part of this kind of circles, you must have experienced this kind of experience at least a little bit, right?

The bit I am interested in is the claim of controlled listening tests showing that poor measuring speakers are preferred to great measuring speakers.

 

[Keith_W]

The bit I am interested in is the claim of controlled listening tests showing that poor measuring speakers are preferred to great measuring speakers.

Well that's a bit of a circular argument, because the definition of a "good measuring speaker" is one that has been verified by preference studies. Otherwise, how would we know that a falling frequency response is preferred to a flat frequency response?

So ... if a controlled listening test shows that a "poor measuring loudspeaker" is preferred, then that "poor measuring loudspeaker" is now a "good measuring loudspeaker". I am sure you know what I mean.

BTW, a lot of horns have dynamic enhancement (i.e. command 90dB, get 92dB). As far as I am aware of, there are NO preference studies that show that this is preferred over a more linear loudspeaker. Since it is nonlinear behaviour, we say that it is "poor measuring". However, a lot of people (myself included) like a bit of dynamic enhancement. What would you say if a study showed that this kind of sound is preferred? Come to think of it, I know of a VST that can do exactly this. Time for me to do another informal blind study (similar to this one) I think!!

 

[tmuikku]

BTW, a lot of horns have dynamic enhancement (i.e. command 90dB, get 92dB). As far as I am aware of, there are NO preference studies that show that this is preferred over a more linear loudspeaker. Since it is nonlinear behaviour, we say that it is "poor measuring". However, a lot of people (myself included) like a bit of dynamic enhancement. What would you say if a study showed that this kind of sound is preferred? Come to think of it, I know of a VST that can do exactly this. Time for me to do another informal blind study (similar to this one) I think!!

Hi, do you have any reference for this? Never heard about it.

 

[robh36062]

BTW, a lot of horns have dynamic enhancement (i.e. command 90dB, get 92dB). As far as I am aware of, there are NO preference studies that show that this is preferred over a more linear loudspeaker. Since it is nonlinear behaviour, we say that it is "poor measuring". However, a lot of people (myself included) like a bit of dynamic enhancement. What would you say if a study showed that this kind of sound is preferred? Come to think of it, I know of a VST that can do exactly this. Time for me to do another informal blind study (similar to this one) I think!!

If it's narrow band sounds like honk to me? That's coloration and I have never experienced that with the horns I have used. In a typical 2 way 15 you would be leaving the woofer behind in a rather glaring way as you completely unbalance the speaker.

Rob

 

[Newman]

Well that's a bit of a circular argument, because the definition of a "good measuring speaker" is one that has been verified by preference studies.

...which by no coincidence happens to be a smooth flat extended FR of the direct-arrival sound above the frequencies where the room is the dominant factor.

Otherwise, how would we know that a falling frequency response is preferred to a flat frequency response?

No, you are measuring the wrong thing. Flat direct sound is preferred. Falling summed sound is a consequence of flat direct sound, if you happen to be using speakers with falling off-axis treble and in a room with falling off-axis treble. If those two secondary conditions are not met, falling summed FR is not preferred.

So ... if a controlled listening test shows that a "poor measuring loudspeaker" is preferred, then that "poor measuring loudspeaker" is now a "good measuring loudspeaker". I am sure you know what I mean.

Yeah but it doesn't happen. Your argument is only circular because your hypothetical isn't real. It is, like I said, no coincidence that the frequency response that sounds like real life is the preferred one. And that isn't a circular argument.

cheers

 

[Duke]

BTW, a lot of horns have dynamic enhancement (i.e. command 90dB, get 92dB). As far as I am aware of, there are NO preference studies that show that this is preferred over a more linear loudspeaker. Since it is nonlinear behaviour, we say that it is "poor measuring". However, a lot of people (myself included) like a bit of dynamic enhancement. What would you say if a study showed that this kind of sound is preferred? Come to think of it, I know of a VST that can do exactly this. Time for me to do another informal blind study (similar to this one) I think!!

I have heard of this from a recording engineer who asked me not to use his name. For decades, inspired by a conversation with Paul Klipsh, in his studio he has been taking measurements of various speakers he had access to which generally (but not always) show a correlation between higher efficiency and reduced compression of peaks. He said he has measured what might be called "dynamic enhancement" with large, very high efficiency speakers. He asked me to keep some of the specifics to myself, as he intends to publish a paper on the subject, but I have no idea where that stands. The Klipschorns were among the speakers he said exhibited measureable dynamic enhancement.

I've not seen the data so can't vouch for anything beyond the fact that the above is what he told me, nor do I have a hypothesis that might explain it.

... having been following Marcel's thread in diyaudio.com: loading and pattern control can be adjusted separately to some extent,

Agreed.

 

[Duke]

Flat direct sound is preferred. Falling summed sound is a consequence of flat direct sound,

My understanding is that the perceived tonal balance is normally a weighted average of the direct sound and the reflections, with the exact weighting probably varying somewhat with the specific conditions (presumably including the direct-to-reflected sound ratio; the frequency response variance between the direct sound and the summed reflections; arrival times of the reflections; and decay times of the reflections across the spectrum).

Is this, or something similar, your understanding?

if you happen to be using speakers with falling off-axis treble and in a room with falling off-axis treble. If those two secondary conditions are not met, falling summed FR is not preferred.

I'm not familiar with situations where falling summed frequency response is not preferred... the implication presumably being that flat or rising summed frequency response is preferred when the two secondary conditions (falling off-axis treble and in-room falling off-axis treble) are not met.

Could you elaborate on what sort of conditions would result in a falling summed frequency response not being preferred, and/or correct me if I've misunderstood you?