• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Effect of Loudspeaker Directivity Compared with In-room Measurements

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
I did a short article on how loudspeaker directivity affects in-room measurements, here is a brief summary:

We know loudspeakers sound different, we also know they can measure different – both in the lab, and in-room. But what if they seem to measure quite similar – and still sound very different. Is there something hidden in those measurements, that we missed?

Frequency response is quite similar:

directivity fr.png


But they sound different. And indeed. if we look deeper, we can find differences in the measurements. Comparing decay profiles reveal smoother more linear decay with controlled directivity:

directivity decay.png


Link to full article:
Effect of Loudspeaker Directivity Compared with In-room Measurements
 

thewas

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
4,317
Likes
9,594
Fully agree that just a single FR doesn't tell much, on the other hand the 2 loudspeakers you show as an example have quite different responses (best seen with higher smoothing like var or psy) so even those alone would be enough for quite different sound.

Here are 2 different loudspeakers which I owned with a surprisingly similar response at the LP that sounded though more different thant someone would have though from that measurement due to their different directivity as you say:

1616149764268.png
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
the 2 loudspeakers you show as an example have quite different responses (best seen with higher smoothing like var or psy) so even those alone would be enough for quite different sound.

Frequency response at lp is different, yes, and this difference is clearly above audible limits. But they would not necessarily be more equal if one tries to eq to same target. Due to differences in decay profile, it is impossible to eq those 2 to achieve identical sound.

Radiation pattern affects what happens in time domain, and those differences have much larger effect on sound than the differences in frequency response alone.

In-room frequency response is also affected by radiation pattern, so that both speakers may very well measure quite similar on-axis, but still show different responses in-room.
 

thewas

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
4,317
Likes
9,594
But they would not necessarily be more equal if one tries to eq to same target. Due to differences in decay profile, it is impossible to eq those 2 to achieve identical sound.
Didn't say anything else, only that imo the example you chose wasn't so optimal as there is an additional unknown variable. A better approach would have been for example to equalise 2 loudspeakers with different directivity to identical anechoic on-axis (or listening window) FR and then measure their in room FR which of course would be different. Or the opposite, EQ them to the same FR at the LP and show that the direct sound (and thus also perceived tonality above transition frequency) is different.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
Didn't say anything else, only that imo the example you chose wasn't so optimal as there is an additional unknown variable. A better approach would have been for example to equalise 2 loudspeakers with different directivity to identical anechoic on-axis (or listening window) FR and then measure their in room FR which of course would be different. Or the opposite, EQ them to the same FR at the LP and show that the direct sound (and thus also perceived tonality above transition frequency) is different.

You can normalize to remove differences in full-window frequency response. It is not necessary to do any eq before measurements are done, it can all be done after, in software.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
The purpose here is to show that there is more information in measurements than what is usually focused on, and that this information can relate to how we perceive sound quality. And in a way that relates to real-world speakers and rooms.

The differences in frequency response (full window) are small, compared to what happens in the time domain. Perhaps a different presentation of the measurements can show this better.

I will look into that.
 

thewas

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
4,317
Likes
9,594
You can normalize to remove differences in full-window frequency response. It is not necessary to do any eq before measurements are done, it can all be done after, in software.
Of course, EQing them to anechoic/windowed measurements would just make additionally sure that the direct sound is really flat.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
I made several pictures with nice or awful graphs, depends on how you see it. I will show one here, a gif animation of trad speaker and CD speaker decay 40ms lines, where the full-window normal line on top is omitted:
directivity decay 40ms.gif

I have plenty more graphs, but I think this will do for now.

We observe several differences between those speakers, and can see that improved radiation pattern affects what we measure:

* Faster very early decay - this is better.
* Decay is smooth as function of frequency - a smooth, tilted curve vs chaos.
* Late reflection level is not very different in level.
* Initial start is more even as a function of frequency, see spectrogram.
 

Duke

Addicted to Fun and Learning
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
907
Likes
2,113
Location
Princeton, Texas
... they sound different. And indeed. if we look deeper, we can find differences in the measurements. Comparing decay profiles reveal smoother more linear decay with controlled directivity...

Very interesting!

I had to eyeball the curves in your first post for a little while to wrap my head around what I was seeing. It is not obvious to me that the first-arrival sound (the loudest, brightest-red curve) of one speaker is better than the other (ignoring the very low bass region). BUT eyeballing the NEXT curve (second-loudest, second-brightest-red), the one on the right is MUCH smoother.

We observe several differences between those speakers, and can see that improved radiation pattern affects what we measure:

* Faster very early decay - this is better.
* Decay is smooth as function of frequency - a smooth, tilted curve vs chaos.
* Late reflection level is not very different in level.
* Initial start is more even as a function of frequency, see spectrogram.

Imo those are ALL very worthwhile accomplishments.

I recall reading about a study where listeners judged the relative loudness of short-duration tones. The tones were all actually the same SPL, but their durations were different. Listeners judged the tones which lasted a little bit longer to be LOUDER. From this, one might expect the speaker on the right [edit: should be left] to sound LOUDER in the 1.5 kHz to 4 kHz region. Is this consistent with your observations?

By the way, VERY NICE job of getting good radiation pattern behavior down all the way down to maybe 250 Hz or so for the speaker on the right. My guess is that there is more to your midbass enclosure than is obvious at a glance.

If I may be so bold, it looks to me like you are getting the BASICS right to an extraordinary extent. This may not be nearly as "sexy" as using exotic materials, but imo it is actually FAR more difficult, rare, and beneficial than using exotic materials.
 
Last edited:
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
I had to eyeball the curves in your first post for a little while to wrap my head around what I was seeing. It is not obvious to me that the first-arrival sound (the loudest, brightest-red curve) of one speaker is better than the other (ignoring the very low bass region). BUT eyeballing the NEXT curve (second-loudest, second-brightest-red), the one on the right is MUCH smoother.

Nothing is obvious until you see it. Even an expert needs to identify and understand what the graphs actually show, and my description is not very extensive, more like absent. I have tried to make this a quick and easy read, for those who are into the tech side of audio.

What we see in the decay plot is the frequency response changing as the window is moved in time, in 40ms intervals, with a pre-window of 40ms.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
From this, one might expect the speaker on the right to sound LOUDER in the 1.5 kHz to 4 kHz region. Is this consistent with your observations?

You mean the left? I really can not say, as it was a year ago that I listened briefly to this speaker in this system, now we had those speakers put away and the others placed in the same location.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
By the way, VERY NICE job of getting good radiation pattern behavior down all the way down to maybe 250 Hz or so for the speaker on the right. My guess is that there is more to your midbass enclosure than is obvious at a glance.

The main speaker has controlled directivity all the way down to 120hz where the bass-system takes over, but the pattern widens at the low end below around 300hz down to 100hz. This speaker measures quite nice, yes, but the goal here was not to show off this marvel, rater than trying to visualize differences in measurements beyond frequency response as it is usually displayed.

And I thank you for your kind words.
 

Duke

Addicted to Fun and Learning
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
907
Likes
2,113
Location
Princeton, Texas
You mean the left? I really can not say, as it was a year ago that I listened briefly to this speaker in this system, now we had those speakers put away and the others placed in the same location.

Oops, yes, I meant the one of the LEFT.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
Oops, yes, I meant the one of the LEFT.

I can add that I remember this speaker to be well-behaved, no ear-piercing highs, no obvious colorations. The measurements also show a well-behaved, well designed speaker, with no obvious flaws. Like you should expect in a top-of-the-line speaker from a well reputed brand, at this price level.

Today, when we have Sterophile with Atkinsons measurements readily available on-line, and now also ASR/Amir, + this new guy, Erin, we can find lots of information on many speakers. Then we also see that a fairly neutral and flat frequency response is not something we can take for granted.
 

tuga

Major Contributor
Forum Donor
Joined
Feb 5, 2020
Messages
3,254
Likes
3,285
Location
Oxford, England
We observe several differences between those speakers, and can see that improved radiation pattern affects what we measure:

* Faster very early decay - this is better.

Can you tell for sure if the faster decay is the result of speaker/room interaction or just a characteristic of that particular speaker?
Do you have free-field measurements to compare with the in-room response?
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
Can you tell for sure if the faster decay is the result of speaker/room interaction or just a characteristic of that particular speaker?
Do you have free-field measurements to compare with the in-room response?

I have quite extensive data for the controlled-radiation speaker, no such data for the trad speaker.

But we can make some quick assumptions based on general knowledge about speakers and how they work; No loudspeaker usable for hifi will have decay rates that comes close to even the most damped room.

If we look closer, we find that after a very short period of time, all sound after a few ms is reflected from the room. In most measurements there are reflections from object near the mic that after only a very short time corrupts the measurement.

I can find IR/ETC from both anechoic and in-room, there we can see that this is the case.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
What we see depends on how we measure, the equipment and how we analyze.

The IR for the normal speaker, from the in-room measurement:
directivity IR normal ht.png


IR of the controlled radiation speaker:
directivity IR F205.png


The peak around 2ms is caused by reflections from mic bow and listening chair. So those measurements are not valid much past 1ms, if the purpose was to analyze the speaker itself only.

In a different room, I did a better job of damping the mic and seating reflections:
directivity IR F205 moderatht.png


Measured closer to the speaker at 1m:
directivity IR F205 1m inroom.png


Now we are getting somewhere. What if we measure in non-reflective environment, can we really get rid of all reflected and late energy that is not coming directly from the speaker?
directivity IR F205 1m nr.png


Gets better and better, but, what are we actually looking at.

Even this last IR is not accurate after around 1-2ms. The level quickly drops so low, that it becomes very difficult to achieve sufficient attenuation of all reflections around mic and speaker to avoid contamination of the measurement.

Earlier than 500us, it is likely that all sound comes from the speaker. If the mic itself is good enough, what we see now depends on how we analyze. Different numerical methods and ways to display the information changes what we see, and how we perceive the measurement.

The IR is useful only for looking at decay and reflections at high frequencies from a speaker. A 500us window is also too short for lower frequencies.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Manufacturer
Joined
Apr 25, 2019
Messages
718
Likes
1,311
Location
Norway
We can look at decay measured closer to the speaker, this should reduce room reflection level.

40ms decay lines, in-room, at 1m:
directivity decay F205 40ms 1m inroom.png


Different room, different position, same distance 1m, same speaker - does it change?
directivity decay F205 40ms 1m inroom 2.png

Does not change, part from the bass range due to position and very different bass-system and location.

40ms is a looooong time. What if we decrease line interval to 10ms:
directivity decay F205 10ms 1m inroom 2.png

Still looks dead where the sound start - from the speaker.

10ms is too long, let's try 2ms:
directivity decay F205 2ms 1m inroom 2.png

Still dead speaker. 1ms - as low as this software can go:
directivity decay F205 1ms 1m inroom 2.png


Decay level rises at lower frequencies with this short window. At higher frequencies, the speaker itself is dead.

This was meant to show that the speaker is much more dead than the speaker-room combination - all late energy in the in-room measurements comes from reflections inside the room.

BUT do not believe a loudspeaker is always completely dead with no audible late energy radiation and no resonances. This is very far from reality. It is just that those resonances and delayed energy levels are so low compared to the room reflections, that they are completely masked in our in-room measurements.

-------
Pictures are too large, we can live with that.
 

hardisj

Major Contributor
Reviewer
Joined
Jul 18, 2019
Messages
2,907
Likes
13,605
Location
North Alabama
this new guy, Erin

LOL ... man, oh, man...

Just FWIW, I have been testing drive units and designing my own speakers for the last 10-11 years. All of the data has been made publicly available. Using Klippel gear, no less. Here's an example of a transducer test from 2011.

I've tested a lot of speakers over the years but, I didn't start publishing anechoic loudspeaker data until around May of last year. And I didn't get the NFS until recently.


I just wanted to clear things up, though, so people don't think I'm new to this. I've been at this for a good while now. :)
 
Last edited:

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,589
Likes
1,349
Also, what about cabinet vibrations and the effect on decay?
BUT do not believe a loudspeaker is always completely dead with no audible late energy radiation and no resonances. This is very far from reality. It is just that those resonances and delayed energy levels are so low compared to the room reflections, that they are completely masked in our in-room measurements.

Is this true with lively cabinets? I would think decay from cabinet vibrations would be clearly audible with some speakers.
 
Top Bottom