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

[Newman]

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?

My best understanding of the research findings is that the direct sound has primary importance for perception of tonal balance. Get that wrong and one is behind the eight ball, almost futile to try and compensate with reflections.

Once the direct sound is right, then getting the reflected sound right will be preferred to getting it wrong. And 'right' means either constant directivity or smoothly varying directivity with frequency.

One could assess the reflected sound from examining the off-axis measurements alone, or in combination with (known) direct sound.

And of course, getting the bass right (which is all reflections), and adjusted in level 'to taste', will be important to perceived tonal balance.

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?

I'll try, but bear in mind, what I am trying to say is that falling summed FR is not automatically preferred (I think Keith stated to be axiomatic, "falling FR is preferred"), and that looking at summed FR is a clunky proxy way to assess the preferred-ness of a frequency response, and can get you into trouble when looking at non-standard gear or rooms. Did someone say horns?

One example: live rooms vs dead rooms. A really seriously dead room (which I have seen quite a few audiophiles aim for) will be best served by an effectively flat 'summed' frequency response, because it's all direct sound energy and direct sound sounds best when flat.

Another example: in-wall speakers in very live rooms. It might take quite a few drivers, but you could put together an in-wall speaker that is close to 180° beamwidth at all frequencies. Put that speaker (or an omni speaker with flat FR) in a very live room with hard furnishings (hello modern decor), and neither the speakers nor the room will be causing much off-axis energy decay that varies with frequency. So, the assumption behind the convention for falling summed FR, is not applicable here either.

None of which is to deny the clear experimental finding that, with standard speakers in standard rooms, a listener preference test will reveal the desirability of a falling summed FR.

cheers

 

[Elgrosso]

For the benefit of other readers, instead of using a flown array you have a horn generating the same effect with a ceiling mounted speaker which is much lower profile.
View attachment 389863

https://jblpro.com/en-US/site_elements/4660-information

For this, the people further away got more of the sound than the people close to the speaker. Then it was balanced for the audience.

When you use this horn horizontally, if you are dead center you get a nice phantom center. Now, the person to the right of you gets a louder left and quieter right experience, so the image of the phantom center is still natural in their mind.

The woofers don’t have this trick, so they will get a little more right woofer because they are closer.

The measurements show that the asymmetry isn’t 100% constant for all frequencies.

Lots of reasons this SHOULDN’T work, but somehow it does!

I used this for a theater with a large 15 ft wide sectional sofa where the people at the edge of the sofa are almost on axis with the left or right speakers. With this layout, it was incredible to have the S2600 work its magic.

In my current theater, the width isn’t as extreme so a more conventional setup works. This is where I use the Meyer Sound Amie.

Note that the bass already starts rolling off as early as 100 Hz for the 12” S/2600. But the actual cabinet has the same horn as the Everest and S/3100 so the footprint is ultra wide (even though it’s not too deep).
...

These asymmetrical horns flipped on their sides remind me the car audio horns concept, like for example from Eric Stevens.
They created a very strong center for all seats even without dsp, directing more sound to the opposite side.

Plus, if well installed, they would take advantage of the dashboard lower shape to virtually increase the mouth size, for more vertical directivity.
I still don't have a clear picture of their global radiation behaviour but that’s what I remember, and it worked very well (for the highly difficult car environment).

 

[Duke]

Interesting...although "bright" could mean (a) "brighter than normal speakers" (after all, you do mention normal speakers as a reference), or could mean (b) "brighter than neutral sounding". I agree with you that (a) could be true, but note that (b) could simultaneously be untrue. It would need to be separately tested.

Good point, and I'll try to convey what I mean with a bit less ambiguity later in this post.

I am going to replicate [Earl Geddes' graphs] here, because I have a couple of comments.
View attachment 389905

© Gedlee LLC, fair use claimed​

The black line is the 20 degree off-axis curve.

That is actually the black line. The red line is the sound power curve.

I think not, otherwise the red would be above the black at 20 Hz.

YOU ARE ABSOLUTELY RIGHT! Thank you for finding and correcting my error! I went back and edited that post.

Here are a few quotes from Earl wherein he says what I was trying to communicate:

"Basically to be accurate the response must be smooth and "near flat." The directivity index (DI) will play into how "flat" it should be as well as the room itself. A higher DI will want more HF roll-off, but it must always be smooth. The DI should always be near flat as possible over as wide a range as possible."

"Directivity (DI) also matters, a lot, but it is still not as important as the listening axis frequency response".

"I think that people have almost universally found that a wide directivity at the HFs will sound bright if this response is flat. Turn it down a bit and it's not so bright. A directive speaker like a dome tweeter can probably get away with a flatter on-axis response because its power response has fallen so much."

(I'm not so sure about the "people have almost universally found" part, but the rest of that quote is the point I was trying to make.)

Those kinds of rolloffs, in the top octave or so, are very common in the measurements (I don't mean ASR measurements, I mean taking a variety of real-world examples) seen from all kinds of speakers, including conventional. I tend to grit my teeth and say, "I'm not going to interpret that, because it could be almost anything: microphone performance, microphone positioning, anti-aliasing filter, air absorption, or the tweeter itself." I would want to be intimately familiar with every detail of the measurement test itself, before drawing conclusions.

I was present when he made a similar set of measurement; my high-tech job was to rotate the turntable to the next horizontal angle for each sweep (and in doing so I learned how to do "spins"). I'm pretty sure the roll-off was "the tweeter itself".

Having raised all my caveats, my best guess is that Earl is ceasing to equalise for the top octave, and allowing the horn's natural rolloff to occur. You can see from the polar plot that the horn's directivity control is collapsing above 12-13 kHz, and we all know that makes it less amenable to EQ in that range without audible negative consequences. Earl would be aware of that.

Agreed.

[Genelecs] are generally dead flat right out to 20 kHz, and yet have a slightly wider beam (say ± 50-60° vs ± 45°) and maintain that width right out to 20 Khz, whereas the Geddes collapses above 12-13 kHz. As a result, Genelecs have even more HF energy off-axis than the Geddes horns. Do we see a pattern of complaints about excessively bright highs with Genelecs? I haven't noticed it.

I haven't noticed that either. The biggest difference I've experienced between speakers with uniform response out to as high as one can hear vs a collapsing polar is that the former sound tonally correct over a wider area. The latter quickly lose that sense of air outside the sweet spot.

Thank you for bringing this up, as I think Genelecs are an excellent example for both of us to make our points.

Here is a graph of the 8030 which shows the on-axis response and the power response, note the downward-sloping line drawn through the power response:

At the risk of oversimplifying (or worse), to a first approximation, the frequency response that the listener perceives will be a weighted average of the direct sound and the power response. I don't know what the weighting would be (Toole and/or Olive have sorted out the subjective influence of the reflections in much more detail than the above graph enables), but for the sake of this discussion let's say it's "two parts direct sound and one part power response". The direct sound is "flat" and the power response trend line slopes down by approximately 9 dB from 20 Hz to 20 kHz, or 3 dB per decade. Our hypothetical "2:1 weighted average" would predict that the perceived frequency response slopes down 3 dB across the audible band, or about 1 dB per decade.

I have no proof, but I THINK this (admittedly hypothetical) 1 dB per decade downward slope in the perceived frequency response is what prevents the Genelecs from sounding bright. I speculate that if the power response were as flat as the direct sound the Genelecs WOULD sound bright, consistent with the findings of Sean Olive, Bruel & Kjaer, and others about power response preference.

@Newman and @jhaider, let me know your thoughts about the validity of my assumption that, to a first approximation, the perceived frequency response will be a weighted average of the direct sound and power response. Yes I know that the listening window response and the early reflections are also predictive of preference, but they aren't shown in the above graph so I can't readily factor them in, hence my terminology "to a first approximation". The main point I am trying to make is this: While the direct sound is what matters most, the reflections have a substantial influence on the perceived frequency response.

cheers

And cheers to you too!

 

[Duke]

My best understanding of the research findings is that the direct sound has primary importance for perception of tonal balance. Get that wrong and one is behind the eight ball, almost futile to try and compensate with reflections.

I agree with the first sentence, but not necessarily the second. Quoting David Smith, designer of the landmark JBL 4430 studio monitor:

“... we know that flat axial response and falling power response can sound good. Such speakers did best in the early Toole study. This does not preclude a gently rising [axial] response from being counterbalanced by a more strongly falling power response, or a falling axial response being balanced by a flatter off-axis power [response].” (emphasis mine)

... what I am trying to say is that falling summed FR is not automatically preferred (I think Keith stated to be axiomatic, "falling FR is preferred"), and that looking at summed FR is a clunky proxy way to assess the preferred-ness of a frequency response, and can get you into trouble when looking at non-standard gear or rooms. Did someone say horns?

One example: live rooms vs dead rooms. A really seriously dead room (which I have seen quite a few audiophiles aim for) will be best served by an effectively flat 'summed' frequency response, because it's all direct sound energy and direct sound sounds best when flat.

I've never been in a “really seriously dead room” but I have been in several overdamped rooms, and they are not my preference. I can see how "flat summed response" could work in such rooms as long as any on-axis tip-up is mild.

Another example: in-wall speakers in very live rooms. It might take quite a few drivers, but you could put together an in-wall speaker that is close to 180° beamwidth at all frequencies. Put that speaker (or an omni speaker with flat FR) in a very live room with hard furnishings (hello modern decor), and neither the speakers nor the room will be causing much off-axis energy decay that varies with frequency. So, the assumption behind the convention for falling summed FR, is not applicable here either.

My instinct is that both would sound too bright in the room you described based on my understanding of Sean Olive's findings, but I could be wrong.

None of which is to deny the clear experimental finding that, with standard speakers in standard rooms, a listener preference test will reveal the desirability of a falling summed FR.

Agreed.

cheers

Cheers to you too, my friend.

 

[Newman]

@Duke, good points all. I don't want to address everything you raise individually, because I sense that we are 'violently agreeing' (but even that is too strong: maybe 'jostling to agree'), with only minor differences in interpretation.

There is no doubt that a speaker in a highly reflective/empty room sounds brighter than the same speaker in a room with lots of soft furnishings and carpets and curtains. The direct sound hasn't changed, so, like you say, the reflected sound is causing it. Agreed.

But in looking at aggregate frequency response curves, Toole's comments on psychoacoustics come into play: "...humans are wonderfully adaptable: We can usually compensate for things we can measure and for things we can hear while we are moving or while they are changing but fade away once stability is established. It is almost as if when we walk into a room, we hear all the reflections, and this gives us a great deal of information about the acoustical nature of the space. Then, when we sit down, within a very short time the perceptual effects of the reflections are attenuated, some more than others, and we settle in to listen to the sound sources, whatever they may be....
The same sound source—a voice or musical instrument, for example—will take on some of the character of different rooms, and yet they can still be recognized as being the “same” sound sources. Within some range of “normal” rooms, we seem to have a built-in ability to “listen through” a room to attend to even minute details of the sound source. It is an interesting tale.
Reflections have many effects on the perception of sound in rooms; some are interactive with each other, and others are relatively independent."

- Sound Reproduction, first edition, sec 4.3.5, fair use claimed​

The bottom line from the above being that Toole urges caution in using aggregate measures (that work quite well in large spaces) to analyse perception in small rooms. For instance, your example of the Genelec 8030C having a power response of -3dB per decade, and your saying that is the reason it doesn't sound bright. Well, the power response curve of the Gedlee Summa, which we have been discussing, falls at -6dB per decade, twice as fast, plus it has less off-axis energy in the top octave, plus less on-axis energy in the top octave. So, since they both have generally flat FR for the direct sound, and the Genelec doesn't sound bright, then the Gedlee has to sound dull, right? Hmm.....OR, maybe other metrics will be more predictive. I lean to the latter.

Finally, when I said that it is 'almost futile' to compensate with reflections for tonal imbalance in the direct sound, I didn't provide enough context. Let's say we have a speaker with a rising direct sound FR, and we wish to compensate with a falling reflected sound FR. And let's say we can fully adjust the reflected sound, maybe using an anechoic chamber with a whole bunch of speakers all over the walls to simulate reflected sound in a highly adjustable way. So we tweak away until the listener says, "Okay that's the right setting, because if you go any steeper the tonality sounds too dull, and less steep sounds too bright." Job done? Not really, because like Toole says, the listener can still hear 'inside' that overall sound and tell that the direct sound has something wrong with it. The listener would much prefer that the direct sound was flat, and reflected sound then adjusted to sound tonally 'right'. That is what I meant by 'almost futile'.

cheers

 

[Duke]

@Newman, yeah I can imagine you and me walking out of the Genelec room at the audio show, and our conversation would go something like this: "As you just heard, those big Genelecs sounded great because they got the direct sound right!" "No no no, they sounded great because they got the reflections right!"

Cheers

 

[jeffaegrim]

Hi, I don't see anywhere in that article claiming a stable centre sound image when seated off-centre. I do see the words, "someone walking a horizontal line between the speakers would be exposed to a constant sound level", but that is not the same thing.


OK...if the two of you are conflating equal sound level from left and right speaker when sitting off-centre, as meaning the phantom centre is maintained, then I don't think it is right to make that claim.

It is (probably slightly) better than nothing, but it would be overreach to say the phantom centre is stabilised when moving from centre seat to an off-centre seat.

cheers

Hey - thank you for the reply, but may I kindly ask you to elaborate what you're getting at? Was I inaccurate in my usage of any terminology? I was merely describing my impressions, though the Everest white paper seems to speak of trying to achieve exactly that: "[...] The chief design aim in this Project Everest system has been to widen the area over which the listener can perceive accurate stereo imaging. This has been accomplished through precise shaping of the system's polar response, largely by means of a unique defineded coverage high-frequency horn design. [...]"

 

[Newman]

Hi @jeffaegrim, I am merely trying to point out that the Everest 'trick' of sitting off-centre and hearing equal loudness from left and right speakers, is not sufficient to create a stable centre phantom image. It won't work. To achieve that goal, you would have to counter the precedence effect.

Which is not to say it would have no effect at all, it would, but very limited.

cheers

 

[robh36062]

Hi @jeffaegrim, I am merely trying to point out that the Everest 'trick' of sitting off-centre and hearing equal loudness from left and right speakers, is not sufficient to create a stable centre phantom image. It won't work. To achieve that goal, you would have to counter the precedence effect.

Have you read the white paper? It does address your concerns and is worth a look.

Rob

 

[Newman]

Ah, that's better! Unfortunately all the relevant information is missing.

From the little that remains, I would say it's a half-baked solution (uses the wrong Haas data on clicks instead of voice or music, hence treats the solution as a UHF problem addressed only above 4 kHz), but better than nothing. Good on 'em.

cheers

 

[multichannel_guy]

Do wide dispersion horns honk more? Is there such a consensus?

 

[robh36062]

Do wide dispersion horns honk more? Is there such a consensus?

Not in my experience. The ones I have used don't "honk" at all.

Rob

 

[multichannel_guy]

Not in my experience. The ones I have used don't "honk" at all.

Rob

I have been rocking to a 60/40 PA horn that doesn't seem to honk. Recently, i heard a wider dispersion home use audiophile speaker that honked like crazy and was very distracting. Hence, i wondered. Is there any technical reference you may have as to what parameters possibly cause the honk sound?

(not mentioning any brand names for fear that brand name fan boys will come out and become unable to have a reasonable or civil discussion thereafter).

 

[Ktacos]

Probably need to define what honk even means.

 

[Sal1950]

Do wide dispersion horns honk more? Is there such a consensus?

Do wide dispersion tweeters tweet more?

 

[Newman]

Do wide dispersion horns honk more? Is there such a consensus?

My initial thought is more the opposite. What made you suggest it?

 

[multichannel_guy]

My initial thought is more the opposite. What made you suggest it?

My ownership of a PA horn is recent, never had a horn speaker before that, but, the only thing i gathered was that the honky horn i heard at this guy's house was "wide dispersion". I didn't say it honked to hurt his delicate audiophilly feelings and lied, i said it touched my soul deeply instead.

Maybe someone like that Audio Corner Erin guy or similar could chip in, tie some variable, measurement to what causes that type of sound.

I googled a bit and found this one comment on audiokarma.

" the "honk" is the high number of reflections within the horn and it causes spikes and nulls, but the spikes is what makes people hear the honk. It is found in older style horns, Exponential (SP), for example. The ****** in your avitar is an Expo horn. The modern wave guides don't have near as many reflections in the horn and don't sound like "horns"."

Not sure what this guy is saying about spikes and nulls or not?
Any feedback from the science guys here?
Does it have to do with the splines, the horn's geometry itself?

 

[robh36062]

It's not simple unfortunately. For example where the crossover frequency is, what CD is used and so on can all contribute. How resonate the material the horn is made of can also effect things. In some cases it just poor implementation using the horn lower than it was designed for. Horn honk is not tied to any particular dispersion pattern.

I have used mostly modern horns and waveguides and simply not had an issue. My Arrays are 70x70 a pair of 100x100, 120x100 and 120x110. Have also used old school round exponentials with no issues.

Rob

 

[Newman]

My ownership of a PA horn is recent, never had a horn speaker before that, but, the only thing i gathered was that the honky horn i heard at this guy's house was "wide dispersion". I didn't say it honked to hurt his delicate audiophilly feelings and lied, i said it touched my soul deeply instead.

It's okay to actually name the horn in question...horns don't have feelings, and it can't be a Tekton so you are safe from legal threats! Then we can at least discuss (a) whether it really was wide dispersion, and (b) the actual attributes of the horn.

Maybe someone like that Audio Corner Erin guy or similar could chip in, tie some variable, measurement to what causes that type of sound.
I googled a bit and found this one comment on audiokarma. " the "honk" is the high number of reflections within the horn and it causes spikes and nulls, but the spikes is what makes people hear the honk. It is found in older style horns, Exponential (SP), for example. The ****** in your avitar is an Expo horn. The modern wave guides don't have near as many reflections in the horn and don't sound like "horns"."
Not sure what this guy is saying about spikes and nulls or not?
Any feedback from the science guys here?
Does it have to do with the splines, the horn's geometry itself?

The most common reason for horn honk is when the horn is used to play frequencies lower than its design merits. As a result, the too-long wavelengths resonate back and forth between the horn's throat and mouth. This is easy to do accidentally by using a low order HP crossover, which fails to attenuate these too-low frequencies to a sufficient degree. This happens too often in 'exotic horn' designs for 'high end home audiophiles', because such audiophiles are known to believe, however misleadingly, that low-order crossovers 'sound better'.

The other common reason is the mouth of the horn being too sharp to allow the sound waves to transfer from the horn walls to the open air without diffraction at the mouth, and consequent reflections back down the horn. This mechanism comes into play most often at the lowish frequencies in the horn's bandwidth, because high frequencies tend to 'let go' of the horn walls before even reaching the mouth. The classic exponential horn has a sharp mouth, but it still won't honk as long as its bandwidth is limited to sufficiently high frequencies. OTOH a much more rounded mouth treatment would have allowed the same horn to go a bit lower before showing signs of honk.

A simple example of both of the above mechanisms is the hand-held 'loud hailer'. Its mouth size probably only warrants usage above 1000 Hz, yet it is being asked to reproduce the full human vocal range. And its mouth is relatively sharp, hence emphasising diffraction and reflection.

There are other mechanisms, such as the Higher Order Moduli (HOM) resonances that Geddes has discussed extensively, where sound waves bounce from side to side of the horn on their way to the mouth. This is certainly a relevant issue, but the above two mechanisms dominate the perception of 'honk'.

I don't think of honk as a dispersion-specific issue, but a not-unrelated observation would be that shorter horns with higher flare rates are less likely to honk if competently designed, and are naturally easier to make with less-sharp mouth treatments.

cheers

 

[multichannel_guy]

It's okay to actually name the horn in question...horns don't have feelings, and it can't be a Tekton so you are safe from legal threats! Then we can at least discuss (a) whether it really was wide dispersion, and (b) the actual attributes of the horn.


The most common reason for horn honk is when the horn is used to play frequencies lower than its design merits. As a result, the too-long wavelengths resonate back and forth between the horn's throat and mouth. This is easy to do accidentally by using a low order HP crossover, which fails to attenuate these too-low frequencies to a sufficient degree. This happens too often in 'exotic horn' designs for 'high end home audiophiles', because such audiophiles are known to believe, however misleadingly, that low-order crossovers 'sound better'.

The other common reason is the mouth of the horn being too sharp to allow the sound waves to transfer from the horn walls to the open air without diffraction at the mouth, and consequent reflections back down the horn. This mechanism comes into play most often at the lowish frequencies in the horn's bandwidth, because high frequencies tend to 'let go' of the horn walls before even reaching the mouth. The classic exponential horn has a sharp mouth, but it still won't honk as long as its bandwidth is limited to sufficiently high frequencies. OTOH a much more rounded mouth treatment would have allowed the same horn to go a bit lower before showing signs of honk.

A simple example of both of the above mechanisms is the hand-held 'loud hailer'. Its mouth size probably only warrants usage above 1000 Hz, yet it is being asked to reproduce the full human vocal range. And its mouth is relatively sharp, hence emphasising diffraction and reflection.

There are other mechanisms, such as the Higher Order Moduli (HOM) resonances that Geddes has discussed extensively, where sound waves bounce from side to side of the horn on their way to the mouth. This is certainly a relevant issue, but the above two mechanisms dominate the perception of 'honk'.

I don't think of honk as a dispersion-specific issue, but a not-unrelated observation would be that shorter horns with higher flare rates are less likely to honk if competently designed, and are naturally easier to make with less-sharp mouth treatments.

cheers

Thanks a lot for this bro, think i am learning something today...
To summarize and relatively simplistically, the dominant honk variables are a aggressive poor choice of the crossover point and the size itself?
Larger the horn and conservative crossover point = 'lesser probability' of honk?