Constant Beamwidth Transducer (CBT) Speakers

[Hayabusa]

I'm saying that it IS an FIR (the drivers themselves plus the driver radiation patterns, of course)

Confusing, how would you model that? A 3D FIR? how to model the whole behavior, frequency dependent directivity in a FIR?

3D FIR -> 5D FIR?

 

[j_j]

Confusing, how would you model that? A 3D FIR? how to model the whole behavior, frequency dependent directivity in a FIR?

3D FIR -> 5D FIR?

You can break it into several independent parts. Saying more than that would require rather a lot of typing and a bunch of formulae. Check the patent, I think it has some decent information.

 

[Y~BNA]

I'm saying that it IS an FIR (the drivers themselves plus the driver radiation patterns, of course) and that an FIR (with perhaps some response adjustments for the driver), applied to each, to provide proper shading ACROSS FREQUENCY, is a good way to go. It does require amp per driver, of course.

For instance:

https://www.researchgate.net/publication/302855458_Loudspeaker_array_providing_direct_and_indirect_radiation_from_same_set_of_drivers

Yeah, i was a bit slow catching on as well (YOU CAN SAY THAT AGAIN!!). You actually make a very important point, specifically for MY quest, but in general too. We should investigate the FIR filter-like acoustical properties, for lack of a better term, of any shape a physical design may take, before proceeding to build it and to apply any DSP-based filtering (FIR or otherwise). I have to dig into that. Tx for the heads up, it was needed..

 

[j_j]

Yeah, i was a bit slow catching on as well (YOU CAN SAY THAT AGAIN!!). You actually make a very important point, specifically for MY quest, but in general too. We should investigate the FIR filter-like acoustical properties, for lack of a better term, of any shape a physical design may take, before proceeding to build it and to apply any DSP-based filtering (FIR or otherwise). I have to dig into that. Tx for the heads up, it was needed..

Yes, yes, yes, and even for non-arrays this has enormous effects in regard to crossover steepness, etc.

 

[audiofooled]

I'm saying that it IS an FIR (the drivers themselves plus the driver radiation patterns, of course) and that an FIR (with perhaps some response adjustments for the driver), applied to each, to provide proper shading ACROSS FREQUENCY, is a good way to go. It does require amp per driver, of course.

For instance:

https://www.researchgate.net/publication/302855458_Loudspeaker_array_providing_direct_and_indirect_radiation_from_same_set_of_drivers

Hi JJ,

Some questions about loudspeaker array, if you don't mind:

1. Regarding direct sound - What are the advantages in having both planar and spherical wavefront possibilities for the first onset of the signal and how does that affect interchannel summation with regards to the perceived distance and elevation cues?

-Can effectively modifying critical distance have an influence on perceived loudness, creating also proximity effects and close to head localization? (As in cues being perceived perceptually closer, rather than perceptually louder)?

Conversely, can slightly reduced correlation in some frequency band make it sound more distant?

2. Regarding indirect sound - It takes a lot of understanding and then processing in what would actually sound as diffuse field when reproduced in small rooms, taking into account it is not always recorded and mixed properly at the source. What are the potential complications with increased number of channels, as far as aliasing effects?

-Is intrachannel processing described in the patent an effective solution to that, or do fewer channels still reduce the potential headache of spatial cues not sounding as natural?

3. Regarding low frequencies - In such a system having physical limitations in low frequency output due to small drivers, normally requiring help from subwoofers, may limited output in the mains still be sufficient in introducing decorrelation, leading to increased spatial perception, reducing the need for a lot of subs, potentially to just a single one?

I.e. can auditory system still process the leading edges of the low frequency signal separately for mono or stereo summation regardless, and still be able to process spatial cues of what comes later?

- What is actually "later" in terms of milliseconds for low frequencies? Can we even have perceptually direct sound in low frequencies in rooms, provided that we take some care of modes in the listening environment?

I do hope my questions are not being too daft.

 

[j_j]

Hi JJ,

Some questions about loudspeaker array, if you don't mind:

1. Regarding direct sound - What are the advantages in having both planar and spherical wavefront possibilities for the first onset of the signal and how does that affect interchannel summation with regards to the perceived distance and elevation cues?

-Can effectively modifying critical distance have an influence on perceived loudness, creating also proximity effects and close to head localization? (As in cues being perceived perceptually closer, rather than perceptually louder)?

Not so much loudness, but by setting it to a near-field (say 2 meter) focus, you can create an amplified distance sensation, so if you're 4 meters away it sounds like you're 100% in the diffuse field, and at 2 meters, you're fully direct. It's a great effect for some things, you can listen to the same track as you walk toward the speaker arrays, the stage zooms up to meet you very effectively.

Conversely, can slightly reduced correlation in some frequency band make it sound more distant?

Correlation? Yes. No. Arithmetic correlation is not the thing that's interesting though. If you can, come to the talk next weekend (not this weekend) I'm giving, and that will cover a lot of this stuff. So the answer is "decorrelation is farther away' but the correlation of WHAT isn't so simple. There are an infinite (literally) number of ways to decorrelate something, and most of them sound just plain silly.

2. Regarding indirect sound - It takes a lot of understanding and then processing in what would actually sound as diffuse field when reproduced in small rooms, taking into account it is not always recorded and mixed properly at the source. What are the potential complications with increased number of channels, as far as aliasing effects?

Not sure why you'd ask about aliasing. If you mean throwing zeros into the spectrum, yes, but that's the goal, but they have to be very, very VERY complicated in position and shape, and there has to be a lot of variation over short spatial positions.

-Is intrachannel processing described in the patent an effective solution to that, or do fewer channels still reduce the potential headache of spatial cues not sounding as natural?

Yes and no. There is a ****load of discussion around that particular question.
In general, you can 'tile the plane' with 7 speaker positions, roughly lcr ls, rs, lr rr, where ls and rs are just slightly more than 90 degrees from front center.

3. Regarding low frequencies - In such a system having physical limitations in low frequency output due to small drivers, normally requiring help from subwoofers, may limited output in the mains still be sufficient in introducing decorrelation, leading to increased spatial perception, reducing the need for a lot of subs, potentially to just a single one?

Not enough data to answer that, sorry.

I.e. can auditory system still process the leading edges of the low frequency signal separately for mono or stereo summation regardless, and still be able to process spatial cues of what comes later?

- What is actually "later" in terms of milliseconds for low frequencies? Can we even have perceptually direct sound in low frequencies in rooms, provided that we take some care of modes in the listening environment?

In great big rooms, maybe. Sometimes.

I do hope my questions are not being too daft.

No, not daft at all. You've obviously tried some of this stuff and have some understanding.

 

[audiofooled]

Not so much loudness, but by setting it to a near-field (say 2 meter) focus, you can create an amplified distance sensation, so if you're 4 meters away it sounds like you're 100% in the diffuse field, and at 2 meters, you're fully direct. It's a great effect for some things, you can listen to the same track as you walk toward the speaker arrays, the stage zooms up to meet you very effectively.

Correlation? Yes. No. Arithmetic correlation is not the thing that's interesting though. If you can, come to the talk next weekend (not this weekend) I'm giving, and that will cover a lot of this stuff.

So the answer is "decorrelation is farther away' but the correlation of WHAT isn't so simple.

So the answer is "decorrelation is farther away' but the correlation of WHAT isn't so simple.

There are an infinite (literally) number of ways to decorrelate something, and most of them sound just plain silly.


Not sure why you'd ask about aliasing. If you mean throwing zeros into the spectrum, yes, but that's the goal, but they have to be very, very VERY complicated in position and shape, and there has to be a lot of variation over short spatial positions.


Yes and no. There is a ****load of discussion around that particular question.
In general, you can 'tile the plane' with 7 speaker positions, roughly lcr ls, rs, lr rr, where ls and rs are just slightly more than 90 degrees from front center.

Not enough data to answer that, sorry.


In great big rooms, maybe. Sometimes.


No, not daft at all. You've obviously tried some of this stuff and have some understanding.

I can't thank you enough for taking the time in giving me so much to think about.

In general, you can 'tile the plane' with 7 speaker positions, roughly lcr ls, rs, lr rr, where ls and rs are just slightly more than 90 degrees from front center.

Some years ago, in a room I had enough space and no WAF to worry about, I tried 7.1 setup on a Yamaha DSP a1000, cycling through most of the possible setups and settings. I was impressed with spatial quality it had to offer, on some of the right tracks. Alas, only the mains were a line array, center was MTM, and the rest were point sources. Thinking about it, there's no chance that would work. No surprise that nothing sounded natural and center was the first to be excluded.

Not so much loudness, but by setting it to a near-field (say 2 meter) focus, you can create an amplified distance sensation, so if you're 4 meters away it sounds like you're 100% in the diffuse field, and at 2 meters, you're fully direct. It's a great effect for some things, you can listen to the same track as you walk toward the speaker arrays, the stage zooms up to meet you very effectively.

My current setup is at about 2.5m focus, with 4m front to rear wall and 7m left to right. Perceived loudness is consistent throughout the room and I can't run away fast enough to perceive any timbre changes or lose pinpoint on the phantom center. Walking towards is exactly as you describe, very cool effect. The difference is that diffuse field is very track dependent and it can at times sound right from anywhere in the room.

So the answer is "decorrelation is farther away' but the correlation of WHAT isn't so simple.

What I noticed is that image containing some well recorded early spatial information, sounds more distant (more far away than front wall). Something to do with leading edges?
On the other hand, harmonics arriving in phase may lead to proximity effect? My system does it quite convincingly, although I have no idea how it comes about. I'm sure it's every bit complex.

If you can, come to the talk next weekend (not this weekend) I'm giving, and that will cover a lot of this stuff.

Sadly, I'm at CET (UTC+1), also quite possibly be working. I do hope it gets recorded?

3. Regarding low frequencies - In such a system having physical limitations in low frequency output due to small drivers, normally requiring help from subwoofers, may limited output in the mains still be sufficient in introducing decorrelation, leading to increased spatial perception, reducing the need for a lot of subs, potentially to just a single one?

Not enough data to answer that, sorry.

My current setup is reinforced by a single center sub/bass module (real one, though, no small box). I thought there's something to it, looking at the mains inferior output in comparison, when set up right, stereo information containing sub harmonics way down in level is enough for sense of space. Perceived as entire wave clusters shifting at a rate of just a few times a second (both front to back and left to right) on a right track. Not at all intuitive, but perhaps it has something to do with both intrachannel and interchannel summation, also some floor bounce control, where nothing is localizable as source, of course.

In great big rooms, maybe. Sometimes.

I believe you. My room is not great, nor is it big, but can it work for a limited amount of time, in milliseconds, even in small rooms? I mean, as far as detection?

(12ms is about front to rear wall distance where it sums at +6db near the rear wall), but pressure at nearfield is way higher, followed by a bit of "quasi anechoic" gap at the cursor):

 

[Head_Unit]

Hey, what CBTs are actually available to buy? Anything besides the two JBLs? (I would count kits if the cabinets were pre-made, but I think Parts Express isn't selling theirs any more???).

 

[somebodyelse]

Few (if they;'re still available)/ none to buy. Seems like you need to make them yourself now.

 

[Y~BNA]

I'm saying that it IS an FIR (the drivers themselves plus the driver radiation patterns, of course) and that an FIR (with perhaps some response adjustments for the driver), applied to each, to provide proper shading ACROSS FREQUENCY, is a good way to go. It does require amp per driver, of course.

For instance:

https://www.researchgate.net/publication/302855458_Loudspeaker_array_providing_direct_and_indirect_radiation_from_same_set_of_drivers

Hi J.J.,

About this remark of yours (tx for your guidance), checking whether i got your point correctly.

Does it mean we first measure each driver, to see what its native curve is?
Then add by DSP to that curve, in order to land at a desired XO slope?
If yes, what is a good source of info on handling potentially ensuing asymmetrical-order XO's?

Or did you mean other consderations (too)?
If yes, what are those?

Mind you, i'm NOT asking for a tutorial, i'll do the studying myself.
Just asking you what to study (with priorty).

Tx for being you, cheers

 

[j_j]

Consider. An array is points in space, ergo an FIR filter.
Response is rather in space than frequency

 

[Hayabusa]

Consider. An array is points in space, ergo an FIR filter.
Response is rather in space than frequency

Its even an infinite number of FIR filters

 

[j_j]

Its even an infinite number of FIR filters

I would say, rather, each filter is a function rather than a single value. Still the same problem, just more variables.