Effect of Loudspeaker Directivity Compared with In-room Measurements

[Spkrdctr]

I found ringing 20 or 30 dB down--effectively masked in the FR plot--but that persisted after other frequencies had fully decayed. And I do hear them, but as an afterglow, not so much as a primary timbral change, depending on what is being played.

Rick "one may not know what is there until it's not there" Denney

Yes, but I think you are a Tuba player. That in itself says a lot. I played as a complete amateur when I was in high school band. Problem is, I wish I had learned keyboards or guitar. Could have used those skills my entire life. Anyway, if you say you can hear it I will believe you, but if anyone else said it, I would NOT believe them.

But then I'm rarely opinionated!

 

[markus]

And those are the questions worth working on now. In 2022. Using a combination of practical listening experiments, and theory and measurements.

That would be a monumental undertaking. Have you read through all the research that already went into it?

You would need to do bias controlled listening tests with a large number of people in a large number of rooms. Sounds like something that could take years.

 

[rdenney]

Yes, but I think you are a Tuba player. That in itself says a lot. I played as a complete amateur when I was in high school band. Problem is, I wish I had learned keyboards or guitar. Could have used those skills my entire life. Anyway, if you say you can hear it I will believe you, but if anyone else said it, I would NOT believe them.

But then I'm rarely opinionated!

Maybe. But as I wrote above, some of what I'm seeing in these delay charts are explaining some pitch-bending phenomenon in my room that perhaps do require some significant listening experience in that range.

I'm still a complete amateur. But I have played tuba my entire life in groups that could almost always use another player, so I don't think your statement is generally true. It's true that I have no role in the late-night jam sessions at bluegrass festivals, but I can live with that. Playing tuba in school was not, of course, the path to being one of the Cool Kids.

Rick "ejected from that path for a variety of reasons" Denney

 

[Kvalsvoll]

That would be a monumental undertaking. Have you read through all the research that already went into it?

You would need to do bias controlled listening tests with a large number of people in a large number of rooms. Sounds like something that could take years.

Don't need to have all the answers at once. And a lot is already known.

Controlled experiments do not require a large number of people. Many problems can be investigated using simpler methods. And I only need to convince myself. Customers are convinced by marketing mostly, and then some actually listen to the speakers, and some of those again are convinced.

 

[markus]

Don't need to have all the answers at once. And a lot is already known.

Controlled experiments do not require a large number of people. Many problems can be investigated using simpler methods. And I only need to convince myself. Customers are convinced by marketing mostly, and then some actually listen to the speakers, and some of those again are convinced.

Yes, a lot is known but the question remains how exactly it all plays together and how it should (meaningful standards!).
Some food for thought: https://openresearch.surrey.ac.uk/e...ty-upon/99511687202346?institution=44SUR_INST

 

[Kvalsvoll]

Yes, a lot is known but the question remains how exactly it all plays together and how it should (meaningful standards!).
Some food for thought: https://openresearch.surrey.ac.uk/e...ty-upon/99511687202346?institution=44SUR_INST

Thanks, will look at that in-between.

 

[Kvalsvoll]

There has been some interest in measurements that show the difference (controlled/cardioid <-> closed ports) on the same speaker - isolating as much as possible all those difference into just the acoustic radiation pattern differences caused by the cardioid solution.

I have data for this, from the same speaker, in 2 different rooms. The problem is to be able to show this in a way that displays those differences - or show that there are no significant measurable differences.

It is not enough to just show that 2 different measurements are different - because they will be, if you scale and zoom to a high enough resolution, due to measurement inaccuracies and noise.

I may have found 2 ways to display this. There is a difference, it is measurable, it can be seen on the charts.

 

[markus]

Thanks, will look at that in-between.

Not sure that thesis deserves just "a look at it in-between" as it goes into detail what you're specifically trying to answer, including references to previous research.

 

[Kvalsvoll]

Not sure that thesis deserves just "a look at it in-between" as it goes into detail what you're specifically trying to answer, including references to previous research.

I have looked at it, a lot of effort has been put into this, lots of text, unfortunately it does not give any new answers. This is from 2013, a lot has happened since then.

 

[Kvalsvoll]

I have put up measurements comparing the same speaker, with disabled and enabled radiation control. They can be found in the F205-thread on my web page.

Basically, what this shows, is a measurable reduction in early reflected sound, when radiation is more controlled in the 100Hz-1KHz range. The frequency response does not show any large differences. This behavior is similar even if the room has quite good acoustic properties with good boundary absorption down to at least 100Hz.

I may be able to present a controlled listening experiment for this. It is possible that sound samples can be available for download, so those interested can listen and hear at least some of the differences.

 

[markus]

I have looked at it, a lot of effort has been put into this, lots of text, unfortunately it does not give any new answers. This is from 2013, a lot has happened since then.

I must have been sleeping

 

[Kvalsvoll]

How exactly (metrics! numbers!) does direct and indirect sound (result of speaker dispersion and room response) influence our perception of spaciousness, envelopment and especially timbre?
How does it affect localization, clarity?
Does it relate to room size? How?
Can or should it be corrected using DSP or room treatments? To what extend?
What numbers are "right"? Is it even possible to define a target when "audio's circle of confusion" is still in full force?

Bringing up this again, trying to give a better answer to why those questions are relevant.

Seen in a bigger context - many problems with sound reproduction has been solved, to the extent that they are no longer something that needs any further study. Of such are amplifiers, dacs, streaming devices - virtually everything before the signal enters the loudspeaker.

A loudspeaker is much more difficult. They still have audible and measurable deficiencies. And there are different opinions on how they should ideally perform - frequency response, capacity requirements, and - especially radiation pattern.

The sound from a speaker placed in a room will be a result of the speakers radiation pattern and room acoustic properties. And we now know quite a lot about how different characteristics of both the speaker and the room affect sound quality.

I have tried - with my measurements in the article described in post #1 - to show how different radiation pattern results in different decay profiles, and why those differences affect our perception of the sound. The frequency response - as typically presented - does not show the differences, we have to look at what happens in time.

Each of the questions above can be further discussed, opening with what is known.

---------
Starting with Room Size - this one is easy, room size does not affect sound, only acoustic properties do. So a small room can perform like a big room, it just need a little different acoustic treatment.

Some properties can however, be difficult to achieve, depending on room size. A very small space can not achive a long enough reverberant tail to let the reverb from the room add to the sense of space already present in the recording.

Some larger rooms, with opening into other larger spaces, or with strange shapes, can be problematic for bass - you get delayed energy in the bass range, that is very difficult to get rid of with acoustic treatment. Contrary to what many believe, it is easier to get good bass response in a smaller, rectangular shaped room.

 

[rdenney]

Bringing up this again, trying to give a better answer to why those questions are relevant.

Seen in a bigger context - many problems with sound reproduction has been solved, to the extent that they are no longer something that needs any further study. Of such are amplifiers, dacs, streaming devices - virtually everything before the signal enters the loudspeaker.

A loudspeaker is much more difficult. They still have audible and measurable deficiencies. And there are different opinions on how they should ideally perform - frequency response, capacity requirements, and - especially radiation pattern.

The sound from a speaker placed in a room will be a result of the speakers radiation pattern and room acoustic properties. And we now know quite a lot about how different characteristics of both the speaker and the room affect sound quality.

I have tried - with my measurements in the article described in post #1 - to show how different radiation pattern results in different decay profiles, and why those differences affect our perception of the sound. The frequency response - as typically presented - does not show the differences, we have to look at what happens in time.

Each of the questions above can be further discussed, opening with what is known.

---------
Starting with Room Size - this one is easy, room size does not affect sound, only acoustic properties do. So a small room can perform like a big room, it just need a little different acoustic treatment.

Some properties can however, be difficult to achieve, depending on room size. A very small space can not achive a long enough reverberant tail to let the reverb from the room add to the sense of space already present in the recording.

Some larger rooms, with opening into other larger spaces, or with strange shapes, can be problematic for bass - you get delayed energy in the bass range, that is very difficult to get rid of with acoustic treatment. Contrary to what many believe, it is easier to get good bass response in a smaller, rectangular shaped room.

Absolutely! Exactly my situation.

I don’t want my space to have an RTA60 of 3 seconds for playback, though I love it when performing music. The recording already includes room effects. What I want is a playback system that plays the recorded room in a way that sounds real in my room. So some reflection control seems like a good idea to me.

The WAF is unfavorable, however.

Rick “whose room isn’t that bad in the scheme of things” Denney

 

[Holmz]

Article is now updated with better graphs, with scaling that shows the differences better. Especially the spectrum plot with 25dB range shows differences much better.


Trad speaker spectrum:
View attachment 171421

F205 spectrum:
View attachment 171422

When looking at this, it makes sense that there is a very audible difference in sound quality.

I also want to repeat my repeating message on people going blind on frequency response charts - here, the frequency response of those speakers are quite similar - similar enough that they should sound if not similar, at least not very different - but, the spectrum now shows why they sound very different; The F205 has a much cleaner transient response and less early-decay "noise".

Of those plots are created more with continuous tones, then they can destructively cancel/sum.
The direct sound, from say a broadband impulse, would have an impulse in the time domain and reflections combing out.

 

[markus]

Starting with Room Size - this one is easy, room size does not affect sound, only acoustic properties do. So a small room can perform like a big room, it just need a little different acoustic treatment.

Not sure I understand what you mean when saying "room size does not affect sound". Guess you would need to define "sound" first.
Room size affects spectrum, angle, delay and level of reflections and therefore the perceived sound.

Some properties can however, be difficult to achieve, depending on room size. A very small space can not achive a long enough reverberant tail to let the reverb from the room add to the sense of space already present in the recording.

Multichannel to the rescue.

Some larger rooms, with opening into other larger spaces, or with strange shapes, can be problematic for bass - you get delayed energy in the bass range, that is very difficult to get rid of with acoustic treatment. Contrary to what many believe, it is easier to get good bass response in a smaller, rectangular shaped room.

This depends on source and listener position. Rooms vary a lot in their acoustical properties so without doing actual measurements it's virtually impossible to get good bass, especially within the critical transition region where subs and mains overlap. Generally stud wall construction as commonly found in the US tends to soften modal effects dramatically whereas in the EU walls tend to be more rigid and often textbook like modal peaks can be observed.

 

[markus]

Yes, a lot is known but the question remains how exactly it all plays together and how it should (meaningful standards!).
Some food for thought: https://openresearch.surrey.ac.uk/e...ty-upon/99511687202346?institution=44SUR_INST

For those who are too lazy to read the whole thesis here's the conclusion with some interesting findings:

• An auralisation system is suitable to use for controlled studies of directivity effects (where it is necessary for multiple loudspeaker directivities, environments and listening positions to be compared directly)
• Theory suggests that the effects of different directivities, combined with different boundary absorption, boundary separation and listener positions, upon the perception of reproduction should be subtle. Original research carried out here using an auralisation system to present simulations of different combinations of these parameters shows that all parameters have an audible and distinguishable effect: changes in loudspeaker directivity, boundary absorption, boundary separation and listener position, can each cause changes in the percep tion of sound reproduction.
• Using this system, directivity is found to be the most influential factor when considering just two reflective boundaries. It becomes less influential in the presence of more reflections (shown to occur experimentally with four/six reflective boundaries), where absorption is the most influential factor. This decrease in the influence of directivity with increased reflections is consistent with previous studies based on the effects of single-reflections. The magnitude of effects caused by different loudspeaker directivities is reduced with the increased presence of reflections.
• Using this system, although speech is found to highlight the effects of different absorption types best, classical music is found to highlight the effects of different directivities most effectively. This is also consistent with previous literature, which suggest that continuous sounds reduced the perceptual thresholds for single reflection effects. Classical music is an effective signal to highlight the perceptual differences between different loudspeaker directivities.
• Boundary separation has less effect on the perception of sound repro duction in comparison to directivity and wall absorption.

 

[Kvalsvoll]

Of those plots are created more with continuous tones, then they can destructively cancel/sum.
The direct sound, from say a broadband impulse, would have an impulse in the time domain and reflections combing out.

Not quite sure what you mean here, but cancellations due to reflections are part of the system being measured, and thus should also be present in the measurement.

How the decaying sound looks like depends on the numeric processing used for analysis - if there are no resonances, the decay will not have ridges or combing, but it may look that way if the window used in processing is too long on the left side, so that each successive slice outwards in time remembers too much of what happened earlier in time.

Such measurements are done using a frequency tone sweep, this is the method that gives the best signal-to-noise ratio, and the data can afterwards be processed to look at it in many different ways, both in the frequency domain and in the time domain. The measurement itself does not introduce errors, it is quite accurate, it is even possible to analyze non-linear behavior.

 

[Kvalsvoll]

Not sure I understand what you mean when saying "room size does not affect sound". Guess you would need to define "sound" first.
Room size affects spectrum, angle, delay and level of reflections and therefore the perceived sound.

When the room is properly treated, there are no resonances or discrete reflections left, there is no direction on the reverb from the room.

What is more diffcult to achieve is a high enough level of late sound to make it loud enough to contribute - if that is desired. To get a large delay, the sound has to reflect several times between boundaries, and since the boundaries are partially absorptive and diffusive, the sound looses energy on each trip around the room, and soon becomes attenuated to very low levels.

Example waterfall from a small room - no resonances left above around 100Hz:

 

[Holmz]

Not quite sure what you mean here, but cancellations due to reflections are part of the system being measured, and thus should also be present in the measurement.

…

• The direct sound will not have cancellations per se.
• It is solely the reflections that cancel with the direct sound and other reflections causing nulls and peaks.
• And short impulses do not cancel in the time domain, they make a comb.
• to get cancelations one needs more of a continuous signal.

Which sort of means that directional speakers, and more absorbtive rooms, should should much different even when the response curves are identical.

perhaps and analogy makes sense here…
When we pull up to a traffic light, and have the FM radio on, and get just right… then the multi path and direct path cancel out the FM ano the radio does quiet.
So there is actually no measurand that exists when one tries to see what FM signal is present.
But if the FM was a short AM burst, then we would get direct and reflected signal with exactly the same amplitude… some N cycles displaced.

We hear the ensemble of the direct and reflected as marketing up the tone though… so it is important in the room.
But we image moistly with the direct signal.

 

[markus]

When the room is properly treated, there are no resonances or discrete reflections left, there is no direction on the reverb from the room.

What is more diffcult to achieve is a high enough level of late sound to make it loud enough to contribute - if that is desired. To get a large delay, the sound has to reflect several times between boundaries, and since the boundaries are partially absorptive and diffusive, the sound looses energy on each trip around the room, and soon becomes attenuated to very low levels.

Example waterfall from a small room - no resonances left above around 100Hz:
View attachment 172721

I thought you wanted to talk about the effects of room size? Now you're talking about absorption?
What you show looks like an anechoic chamber and also like something where speaker directivity is less of a significant factor? In any case probably not what you'll find in most homes. Can you make that .mdat available for download together with pictures of the room?