Ideal curves for different radiation patterns. Optimize for on-axis or off-axis?

[napilopez]

So one thing I've never quite been able to fully understand is what type of curves we should be looking for when speakers are not the usual monopoles.

For example, we know that on-axis curve should be flat with a 0 slope, while we want the predicted in-room response (usually nearly identical to the early reflections curve) to be sloping down roughly 8-10 dB from 20Hz to 20KHz.

But what do we prioritize for speakers with omnidirectional radiation? Presumably, a perfectly omnidirectional speaker could not achieve both flat on-axis and tilting off axis. Would flat on-axis sound too bright, meaning that the on-axis curve should be tilted down?

But then we also hear the speaker "through the room" so shouldn't we aim for flat? After all, in an anechoic chamber, a flat omnidirectional speaker and a flat monopole on-axis will theoretically sound the same.

It is also something of a debate even among monopole speakers. Is a smooth sloping PIR curve like Revel Salon2 better than the one that tilts down and then straightens out like the M2?

For reference, Revel Salon2:


Vs JBL M2:


We know the Salon2 won in a pair of well-executed blind tests, but I suspect that was likely due to people's general preferences for wide directivity (most people seem to prefer wide directivity, but it is a true matter of preference rather than a quasi-universal thing like flat on-axis).

In general, DR Toole has said:
"Cone/dome loudspeakers tend to show a gently rising directivity index (DI) with frequency, and well designed horn loudspeakers (like the M2) exhibit quite constant DI over their operating frequency range. There is no evidence that either is advantageous - both are highly rated by listeners. "

Furthermore, Sean Olive's ever-important paper on speaker preferences says that it's possible different slopes are needed for different directivities - wider directivity speakers appeared to require a less steep PIR slope. That might be something that's affecting our current use of the preference score.

These are the questions that keep me up at night.

 

[aarons915]

This is an interesting topic, I recall reading in Dr. Toole's Sound Reproduction 3rd edition he cites an experiment by Linkwitz where he compares 2 of his designs, 1 a monopole and the other a Bipole I believe. Both had a very similar ideal room curve and he noted that they sounded very similar timbrally, the differences were in their spatial presentation.

I personally think 1 of the 2 directivities Dr. Toole mentions has to be preferred over the other and various blind tests lead me to believe the gently rising DI is the preferred response.

 

[Blumlein 88]

I agree, it would be my guess an even rising DI is going to be more preferred than a flat one or one with flat sections in it.

I'm also thinking the predictive formula based upon this info, while something never before available, is in need of quite some refinement.

As for the omni vs a more directional speaker, it seems clear the flat off axis response of a flat on-axis omni is going to be somewhat detrimental. So does a slight anechoic slope sound better with an omni. I wouldn't be surprised if that were the case, but I'd think any omni is somewhat disadvantaged as it makes the room reflections more of an issue compared to the directional speakers which make on-axis direct sound more prominent vs room effects above Schroeder.

 

[Soniclife]

It's a question I've wondered about to. The question I bump into quickly is how much does the circle of confusion lock us into liking things the way they have always been. If music was mastered on Omnis would we want a different result.

Are there any in room measurements of a good omni.

 

[napilopez]

This is an interesting topic, I recall reading in Dr. Toole's Sound Reproduction 3rd edition he cites an experiment by Linkwitz where he compares 2 of his designs, 1 a monopole and the other a Bipole I believe. Both had a very similar ideal room curve and he noted that they sounded very similar timbrally, the differences were in their spatial presentation.

I personally think 1 of the 2 directivities Dr. Toole mentions has to be preferred over the other and various blind tests lead me to believe the gently rising DI is the preferred response.

Re-reading that Linkwitz section of the book, in which Toole is largely quoting Linkwitz in agreement, I'm inclined to think it's largely about optimizing for flat on-axis, or close to it that should be the priority. And then let off axis just be close enough to the on axis that it doesn't kill the sense of space. It seems we interpret sufficiently delayed reflected sounds as "room sounds" rather than speaker sounds - so omnidirectional speakers would just make us more aware of our side and rear walls?

Definitely something worth exploring more. Unfortunately its pretty hard to find good data on on omnis.

 

[pozz]

It seems we interpret sufficiently delayed reflected sounds as "room sounds" rather than speaker sounds - so omnidirectional speakers would just make us more aware of our side and rear walls?

When the separation of reflected sounds is within 30ms or so of the direct sound, what would have been distinct localizations if the delay were longer turn into a sense of spaciousness. Omnis take advantage of this property.

 

[hardisj]

I have to think the omni/directed area topic is really based on a) preference and b) the room it's in. For example, I use a set of JBL Pro CDs in a horn that controls directivity down to about 500hz in my home theater because they suckers get downright loud and give you that "I'm in a theater" feel. The speakers literally came from the old theater in town when they closed down. But they're not great for music because I prefer the increased spaciousness a speaker with lower directivity provides. That may not be the most scientific answer and obviously YMMV.

 

[napilopez]

I have to think the omni/directed area topic is really based on a) preference and b) the room it's in. For example, I use a set of JBL Pro CDs in a horn that controls directivity down to about 500hz in my home theater because they suckers get downright loud and give you that "I'm in a theater" feel. The speakers literally came from the old theater in town when they closed down. But they're not great for music because I prefer the increased spaciousness a speaker with lower directivity provides. That may not be the most scientific answer and obviously YMMV.

Thanks for the insight. I definitely think it's a matter of preference and program material with regards to spaciousness, but I guess I'm also largely talking about timbre. What is timbrally balanced for an Omni, flat or a tilt on axis? I'm guessing the former.

 

[Soniclife]

What is timbrally balanced for an Omni, flat or a tilt on axis? I'm guessing the former.

When you say Omni do you mean in every direction, or just horizontally?

 

[Snarfie]

I'm not a specialist by far regarding Room correction software or which target curves should be used. Basically I'm using Room correction software where you can add predefined target curves but better target curves that you can draw such that it suite you so i did. Enclosed setup using a close monitoring an full range setup. Grey line is the found frequency response found in my man-cave. White line is my custom made hand drawn target curve that sounds for me the best. Besides Mathaudio i could not find this feature (using a slider an drawing capability on the fly while listening) on any other room correction software. Maybe there is software that let you do the same if so let me know. Think it would be an good idea to review Room correction software in general the features are quite complex to comprehend.

IMF Compact II monitors in the back.

Upfront Vandersteen Model 1 Full range speakers. Flat response curve suited me the best any other target curve like Harmen did alter specific the highs to much.

The setup:

 

[Jon AA]

I agree, it would be my guess an even rising DI is going to be more preferred than a flat one or one with flat sections in it.

I think that's probably the case, and the shootout between the M2 and Salon2 is a good example of that. However, I think the optimum will change with different applications. A controlled directivity speaker can have many advantages in a theater or multichannel system that won't show up in a 2-channel listening test--in fact they may work against them in such a test.

In a theater/multichannel room where the spaciousness and envelopment in a recording are provided by the surround speakers (so you want/need fewer room reflections), things like dialog intelligibility favor a higher percentage of direct vs reflected sound and you're trying to provide the most similar sound to a wider area/number of listeners, I don't think you end up with the same desired DI curve as you do for a 2-channel room optimized to a single sweetspot.

 

[napilopez]

When you say Omni do you mean in every direction, or just horizontally?

Mostly thinking of a theoretical idea omni, so every direction I suppose.

 

[Soniclife]

Mostly thinking of a theoretical idea omni, so every direction I suppose.

Does such a speaker exist?

 

[ctrl]

We know the Salon2 won in a pair of well-executed blind tests

For me, the decisive difference between M2 and Salon 2, apart from the low frequency reproduction and radiation in general, is the diametrical behaviour with regard to sound radiation in the 1-4kHz range.

With the JBL M2, the "sound energy" of the early reflections increases slightly in the 1-4kHz range, while with the Revel Salon 2 between 2-3kHz it decreases considerably - particularly the sensitive area around the ear canal resonance (approx. 2.7kHz).

For a studio monitor, the behaviour of the JBL M2 is actually ideal, as it prevents overly aggressive mixing.

For a hifi loudspeaker I would apply different standards, since it is important to reproduce as many different audio recordings as possible in a neutral but not too aggressive way - see also "circle of confusion". This is the strength of Salon 2 because of the specific sound radiation at 1-4kHz.

In my opinion the blind test results in the AVS Forum support this thesis, at least they don't contradict it

The typical demonstration song "spanish harlem - rebecca pidgeon" shows a strongly reduced 2-3kHz frequency range in the frequency spectrum - rating almost equal.

With the "battlestar galactica" theme the horns quickly sound unnatural when the range 1-4kHz is emphasized during playback - M2 rated much worse.

The "depeche mode" song with the "cooking lid snare" is immediately annoying if the range 1-4kHz is emphasized during playback - M2 scores significantly worse.
etc.

 

[aarons915]

Does such a speaker exist?

I think a more realistic example is something with a RAAL 64-10, it's not omni but it is very wide dispersion compared to the typical dome. The Philharmonic BMR is a good example of a speaker with wide dispersion, in my small room it makes them a bit bright or unnaturally detailed but many people love them.

 

[test1223]

Hello,

interesting topic. One important part wasn't mentioned jet, the listening distance and the ratio of direct sound to reverberation sound at the listening position. If the listening distance is high the speaker should radiate less omnidirectional. The recommendations of the EBU also suggesting this. From my experience a very good far field setup sounds better than a very good near field setup since the sound is more "enveloping" without becoming "blurred".

A omnidirectional speaker at normal listening distances sounds "enveloping" but "blurred", since the reverberation time (cause by the speaker at the listening position) is to high according to the EBU recommendation. A near field speaker in the near field sounds not "blurred" but also not "enveloping" even if the measured reverberation time is good. A far field speaker in the far field sounds not "blurred" and "enveloping" if the reverberation time meet the EBU recommendations. An exception it an omnidirectional speaker in the very near field in the middle of the room like the Linkwitz Pluto which sounds also not "blurred" and "enveloping".

Best
Thomas

 

[williamwally]

For me, the decisive difference between M2 and Salon 2, apart from the low frequency reproduction and radiation in general, is the diametrical behaviour with regard to sound radiation in the 1-4kHz range.

With the JBL M2, the "sound energy" of the early reflections increases slightly in the 1-4kHz range, while with the Revel Salon 2 between 2-3kHz it decreases considerably - particularly the sensitive area around the ear canal resonance (approx. 2.7kHz).

For a studio monitor, the behaviour of the JBL M2 is actually ideal, as it prevents overly aggressive mixing.

For a hifi loudspeaker I would apply different standards, since it is important to reproduce as many different audio recordings as possible in a neutral but not too aggressive way - see also "circle of confusion". This is the strength of Salon 2 because of the specific sound radiation at 1-4kHz.

In my opinion the blind test results in the AVS Forum support this thesis, at least they don't contradict it
View attachment 48763
The typical demonstration song "spanish harlem - rebecca pidgeon" shows a strongly reduced 2-3kHz frequency range in the frequency spectrum - rating almost equal.

With the "battlestar galactica" theme the horns quickly sound unnatural when the range 1-4kHz is emphasized during playback - M2 rated much worse.

The "depeche mode" song with the "cooking lid snare" is immediately annoying if the range 1-4kHz is emphasized during playback - M2 scores significantly worse.
etc.

Great thread and I'm glad to see people got together to conduct a test like this! Very valuable, but lots of work, kudos to all involved.
Without having to dig through 1,751 posts; do you know how was this test done/setup? Mono, stereo, listening distance, how close to walls, size room, RT etc? Thanks!

 

[ctrl]

Without having to dig through 1,751 posts; do you know how was this test done/setup? Mono, stereo, listening distance, how close to walls, size room, RT etc? Thanks!

As far as I can remember the test was done in a home cinema (showroom) in stereo. A few details can be found here.

 

[Soniclife]

As far as I can remember the test was done in a home cinema (showroom) in stereo. A few details can be found here.

I've always wondered if the result would change if it was done again in a different room, or if room after room the result would hold.

 

[ctrl]

I've always wondered if the result would change if it was done again in a different room, or if room after room the result would hold.

In his book, Toole cites a study by Sean Olive that had three similarly rated speakers in four different rooms evaluated.
The result was that the rating of the speakers among themselves hardly changed.
Quote: Toole, Sound Reproduction

All tests were double blind. In each room, three loudspeakers were evaluated in three locations for each of three programs. The whole process was repeated, resulting in 54 ratings for each of the 20 listeners. These were the results from a statistical perspective:
■ “Loudspeaker” was highly significant: p = 0.05.
■ “Room” was not a significant factor.
■ The results of live and binaural tests were essentially the same.

A possible interpretation is that the listeners became familiar with—adapted to—the room they were in and, this done, were able to accurately judge the relative merits of the loudspeakers.

Since they were given the opportunity to become familiar with each of the four rooms, they were able to arrive at four very similar ratings of the relative qualities of the loudspeakers. Obviously, part of this adaptation, if that is the right description for what is happening, is an accommodation for the different loudspeaker locations.

Different rooms and different positions within those rooms have not confused listeners to the point that they were not able to differentiate between and similarly rate the loudspeakers.