Effect of Loudspeaker Directivity Compared with In-room Measurements

[thorvat]

This method described in the article to achieve consistent imaging across a wide area is only partially successful. I will try to explain why, and it is easy to test for yourself, on your own system.

We have a vocal located in center position:

View attachment 175016

Then we move it to the left, either by moving our listening position to the left, or delaying the right speaker:
View attachment 175017
To restore this, level is reduced on the left speaker and increased on the right speaker. What we get, is something like this:
View attachment 175018

To fully correct for moving LP to the left you would need not only to adjust levels of the left and right speaker but also to adjust the timing by delaying the left speaker so it matches the right speaker. In other words, in order to fully compensate for moving LP to the left, you would need to adjust not only amplitude but also the phase of the left and right signal.

 

[markus]

This method described in the article to achieve consistent imaging across a wide area is only partially successful. I will try to explain why, and it is easy to test for yourself, on your own system.

We have a vocal located in center position:

View attachment 175016

Then we move it to the left, either by moving our listening position to the left, or delaying the right speaker:
View attachment 175017
To restore this, level is reduced on the left speaker and increased on the right speaker. What we get, is something like this:
View attachment 175018
The image moves back to center, sort of, but it is no longer a solid, consistent image, parts of it is still left back on the left side, and the overall impression is a smeared image that lacks the clarity and solidity of the original.

This can be explained from what is known about how hearing works. The higher frequencies are moved back quite well, so the higher frequency spectrum from the object is now moved back to center. The lower frequency part which depends on time arrival for location, is not much affected, and thus remain at its far left location. This gives an image that is now spread around and no longer resembles a solid, connected object with clear boundaries.

In the article, it is claimed that the objects are located in the same position when observed from across the whole listening area. And this can be true, for that specific system and set-up. For a system that is able to reproduce realistic images, which requires control across a much wider frequency range, this is not true - the images will change when you move far off axis, and it is not possible to fix this by manipulating level from the left or light channel speaker.

What can be fixed, is tonality and frequency response errors at far left or right locations, and this will also give a much better spatial impression at those locations. Allowing images to move a little towards the closer speaker is not destructive to the experience.

IID "works" over the whole audible frequency band – see Blauert.
And here comes the interesting part: In stereo interchannel level differences translate to interaural time differences. See the Lipshitz paper mentioned 10 days ago.

 

[Vuki]

To fully correct for moving LP to the left you would need not only to adjust levels of the left and right speaker but also to adjust the timing by delaying the left speaker so it matches the right speaker. In other words, in order to fully compensate for moving LP to the left, you would need to adjust not only amplitude but also the phase of the left and right signal.

Would that also correct reflections arrival time?

 

[markus]

Would that also correct reflections arrival time?

Of course not but does it matter? In order to spread or even shift a phantom image the level of a reflection would need to be pretty loud.

All in all time-intensity trading is a band-aid trying to mitigate an issue inherent to stereo. But I'd rather see speaker directivity shaped in a way that actually solves known problems (small sweet spot, timbre affecting floor/ceiling reflections, etc.pp.) instead of coming up with solutions that obsess about a single specific performance criteria that perceptually might not even be so important. Often other important performance metrics are sacrificed along the way. A cardioid for example sacrifices max. SPL for pattern control. We have had plenty of such solutions in the past yet audiophile discussions from this year read the same as audiophile discussions from decades ago.

 

[sigbergaudio]

We actually don't see any significant reduction in efficiency or increase in distortion in our current prototype cardioid design.

 

[markus]

We actually don't see any significant reduction in efficiency or increase in distortion in our current prototype cardioid design.

Well, that energy used to create the cardioid pattern needs to come from somewhere, no? In the end it's all about the engineering goal and trade-offs. My point was that often the wrong trade-offs are made or the engineering goal itself is ill-defined. I wasn't criticising your designs. Actually I don't know them. So no need to be so defensive all the time.

 

[Vuki]

I still don't understand what's wrong with small sweet spot for serious listening?

 

[thorvat]

Would that also correct reflections arrival time?

How late reflections will come relative to the direct sound is dictated by physical things (speaker position in the room) so the difference in arrival time between direct sound vs reflected sound will always remain the same and cannot be manipulated by DSP.

What it would correct is that direct sound arrival time to the LP of the left and right channel would be the same.

 

[thorvat]

We actually don't see any significant reduction in efficiency or increase in distortion in our current prototype cardioid design.

Well, that highly depends on what you consider "significant", right? I'm joking here..

Of course there is a price to pay for cardioid design, as it is with everything else in life because free lunch indeed doesn't exist. But if it is designed and executed properly benefits will outweight what you loose on other sides, namely the things you mentioned in your post.

 

[markus]

I still don't understand what's wrong with small sweet spot for serious listening?

Nothing but having a perfect sweet spot at the center and slightly compromised phantom imaging at locations next to it is better than having just a good central sweet spot and having images collapse into the nearest speaker at other listening locations.

 

[ahofer]

Nothing but having a perfect sweet spot at the center and slightly compromised phantom imaging at locations next to it is better than having just a good central sweet spot and having images collapse into the nearest speaker at other listening locations.

I feel that my own needs are under-appreciated by the market. I have no trouble staying in the sweetspot horizontally, but I slouch as the listening session goes on.....

 

[markus]

I feel that my own needs are under-appreciated by the market. I have no trouble staying in the sweetspot horizontally, but I slouch as the listening session goes on.....

Headphones

 

[ahofer]

Headphones

Or maybe stands with a remote controlled, or face-activated angle adjustment. And floor bounce compensation...

 

[markus]

To fully correct for moving LP to the left you would need not only to adjust levels of the left and right speaker but also to adjust the timing by delaying the left speaker so it matches the right speaker. In other words, in order to fully compensate for moving LP to the left, you would need to adjust not only amplitude but also the phase of the left and right signal.

No delay is necessary as that's what time-intensity trading helps to overcome. There are other problems with the concept though...

By the way, a similar design as what you suggest was actually proposed by tracking the listener location. As a consequence it would only work for one listener only. Can't find the paper at the moment. Let me know if you're interested in the reference and I'll try to look it up again.

 

[JanRSmit]

To fully correct for moving LP to the left you would need not only to adjust levels of the left and right speaker but also to adjust the timing by delaying the left speaker so it matches the right speaker. In other words, in order to fully compensate for moving LP to the left, you would need to adjust not only amplitude but also the phase of the left and right signal.

Not really, intensity-time trading works and has been researched already long time ago. Be it in a certain band. I need to find an article about it, but i built such systems some 30 years ago, and it worked flawless. The left-right placement can be completely covered between ~400 and ~3500 Hz. Left-right discrimination tails are from some 200 to some 8000 Hz, the maximum effect is around 1300-1600 Hz. We used a 5 inch broad range mid-range angled in so its radiation pattern would give the right intensities at different listening positions.
By the way , same studies indicated that the bass range is highly important in the discrimination of distance. in the solutions we built then, the bass and high frequencies were actually filled in with units both left and right, so 2 woofers, and 2 tweeters.
The spatial image observed was really very nice, full width and depth and discrimination of the sounds in that space (instruments, voices, reverberations).

 

[sigbergaudio]

Well, that energy used to create the cardioid pattern needs to come from somewhere, no? In the end it's all about the engineering goal and trade-offs. My point was that often the wrong trade-offs are made or the engineering goal itself is ill-defined. I wasn't criticising your designs. Actually I don't know them. So no need to be so defensive all the time.

I'm not really being defensive (at least not intentionally), but you are bashing cardioid and/or controlled directivity design without really having any other arguments than that you personally don't believe in them and/or somehow don't feel the result is worth whatever cost is involved. It's getting a bit old.

 

[sigbergaudio]

Well, that highly depends on what you consider "significant", right? I'm joking here..

Of course there is a price to pay for cardioid design, as it is with everything else in life because free lunch indeed doesn't exist. But if it is designed and executed properly benefits will outweight what you loose on other sides, namely the things you mentioned in your post.

Well, we've compensated any potential loss by using a high capacity, high sensitivity midbass 12" driver so there's plenty of headroom available. In practice we've seen a slight increase in distortion at around 100hz, and no significant change anywhere else in the frequency band. And even there we're still at less than 2% THD@100dB/1m at 100hz. And that's before adding the subwoofer that's part of the system and crosses over at 100hz. I guess you can say that the "cost" is that it's overengineered, but I don't feel like we're compromising SPL or anything else.

So if the cardioid is tuned so it reduces existing cancellations and works over a relatively narrow band, the net result can actually be positive. I wrote a bit about it here:

Sigberg Audio Manta (12" wideband cardioid active speakers) development thread

Is your solution somewhat similar to Seas KingRoy Kit or any fundamental differences besides the woofer size I see how it would look similar on the surface, but I would say there are next to no similarities except the use of a coax and somewhat similar cabinet size (Kingroy is smaller). This...

www.audiosciencereview.com

(And my apologies for "advertising" for our own stuff here, but for now it's a development thread and not a commercial product, and I thought it was relevant to the discussion).

 

[thorvat]

No delay is necessary as that's what time-intensity trading helps to overcome. There are other problems with the concept though...

By the way, a similar design as what you suggest was actually proposed by tracking the listener location. As a consequence it would only work for one listener only. Can't find the paper at the moment. Let me know if you're interested in the reference and I'll try to look it up again.

The effect you linked works with sub 1ms time periods, which translates to less than 33cm movements. It won't work that way when you move LP for larger distances toward one speaker, say 1 or 2 meters.

 

[thorvat]

Well, we've compensated any potential loss by using a high capacity, high sensitivity midbass 12" driver so there's plenty of headroom available. In practice we've seen a slight increase in distortion at around 100hz, and no significant change anywhere else in the frequency band. And even there we're still at less than 2% THD@100dB/1m at 100hz. And that's before adding the subwoofer that's part of the system and crosses over at 100hz. I guess you can say that the "cost" is that it's overengineered, but I don't feel like we're compromising SPL or anything else.

So if the cardioid is tuned so it reduces existing cancellations and works over a relatively narrow band, the net result can actually be positive. I wrote a bit about it here:

Sigberg Audio Manta (12" wideband cardioid active speakers) development thread

Is your solution somewhat similar to Seas KingRoy Kit or any fundamental differences besides the woofer size I see how it would look similar on the surface, but I would say there are next to no similarities except the use of a coax and somewhat similar cabinet size (Kingroy is smaller). This...

www.audiosciencereview.com

(And my apologies for "advertising" for our own stuff here, but for now it's a development thread and not a commercial product, and I thought it was relevant to the discussion).

I agree, it seems like you did a fine job there.

 

[markus]

I'm not really being defensive (at least not intentionally), but you are bashing cardioid and/or controlled directivity design without really having any other arguments than that you personally don't believe in them and/or somehow don't feel the result is worth whatever cost is involved. It's getting a bit old.

That explains why you act so defensive. I'm not "bashing cardioid and/or controlled directivity design". I'm just outlining (along with some personal remarks about the current state of affairs in audio land) what needs to be done to show their perceptual benefits. In fact I'm listening to a very high and controlled directivity design for quite some time now. Largely because I have no time to do the testing I'd like to do but also because it works pretty good for me. But never would I jump to conclusions and open a thread presenting the effects of that design as desirable just because some graphs show a rather nebulous difference in late energy in my room. That's just highly unscientific in my opinion.