• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Those of you who believe measurements aren't the whole story, do you have a hypothesis why that is?

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,597
Likes
2,235
I think it's evidence that FR is already left to chance. The curve that hits your ear is the far end of a sequence of chances. The first chance being flat has a random eventual outcome.

Is anechoic not understood here? Trying to throw you a bone...
 

Inner Space

Major Contributor
Forum Donor
Joined
May 18, 2020
Messages
1,285
Likes
2,938
Is anechoic not understood here? Trying to throw you a bone...
First arrival frequency response.....hint hint. @Inner Space

Understood, of course, but we were talking in-room measurements, of the type often displayed here, and obsessively refined and adjusted, until a satisfactory result is achieved - which arithmetically could be achieved just as fast with any FR from the first element.
 

Duke

Major Contributor
Audio Company
Forum Donor
Joined
Apr 22, 2016
Messages
1,523
Likes
3,745
Location
Princeton, Texas
So, you don't really understand the significance of an anechoically flat frequency response?

Apparently not, in your estimation.

Maybe you can tell me what you think I think? I'm not sure how to respond because I don't know what you think I think. It is possible that I communicated poorly, and it is also possible that I don't really understand the significance of anechoically flat frequency response.

In the meantime I'll toss out an idea: Imo there are two things a loudspeaker should get right: The direct sound, and the reflections. Getting the former right is rather straightforward; getting the latter right is a lot more complicated. Let me know your thoughts about this idea, if you don't mind, as maybe we can find some common ground.
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,836
Likes
16,503
Location
Monument, CO
Are you aware of any peer-reviewed research which demonstrates the successful prediction of loudspeaker spatial qualities from measurements? I'm not saying it can't be done, but I'd be very interested in reading about it if it has.

No. Or rather I have a vague memory from years ago of a couple of articles on the subject back when I was taking those acoustics classes in college, but I have zero hope of finding the links now. Consider it the theoretical babblings of a deranged mind. The places I have seen the theory applied in practice were in my day job, working with FLIR, RADAR, LIDAR, and SONAR imaging systems, and I am pretty sure the same techniques would work for a sound field. But I ain't done it, and it would take some work. I think John is right that a pair of speakers in an anechoic chamber would be easier.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,836
Likes
16,503
Location
Monument, CO
You got it all wrong. Imaging in audio terms isn't even about where frequencies combined constructively and destructively. This imaging is a perception created by our brain when we hear the sound.

Quite possibly. Psychoacoustics is not something I am competent to speak about. My thoughts were more along the lines of prior work experience, not in acoustics (except SONAR counts, sort of), mentioned in my response to Duke. After all, the thread asked for a hypothesis, so I gave mine. :)

The posts made me dig up my old post on comb filtering, because the image shifts and change in the tone of instruments as they swept up and down the scale from my Maggies were what led me to my hypothesis. I took some measurements (unpublished and long lost, sorry @Duke) and added treatment behind my Maggies to solve the problem. That was back in the 1980's and perhaps I identified the wrong problem even though the solution worked.

And it is certainly possible that what I think of as imaging does not follow audio terms.
 
Last edited:

Duke

Major Contributor
Audio Company
Forum Donor
Joined
Apr 22, 2016
Messages
1,523
Likes
3,745
Location
Princeton, Texas
The places I have seen the theory applied in practice were in my day job, working with FLIR, RADAR, LIDAR, and SONAR imaging systems...

Omigosh your day job detects, tracks, engages, and destroys my day job!
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,836
Likes
16,503
Location
Monument, CO
Omigosh your day job detects, tracks, engages, and destroys my day job!

Not any more, so you're safe. :) Doing different stuff now, though still high-speed analog.

And some of it was satellite stuff, and work on the space shuttle's radar, so destruction was not always involved.
 

escksu

Addicted to Fun and Learning
Joined
Jul 16, 2020
Messages
965
Likes
397
Maybe. Psychoacoustics is not something I am competent to speak about. My thoughts were more along the lines of prior work experience, not in acoustics (except SONAR counts, sort of), mentioned in my response to Duke.

No. Or rather I have a vague memory from years ago of a couple of articles on the subject back when I was taking those acoustics classes in college, but I have zero hope of finding the links now. Consider it the theoretical babblings of a deranged mind. The places I have seen the theory applied in practice were in my day job, working with FLIR, RADAR, LIDAR, and SONAR imaging systems, and I am pretty sure the same techniques would work for a sound field. But I ain't done it, and it would take some work. I think John is right that a pair of speakers in an anechoic chamber would be easier.

Yes, your job was on different aspect of audio and photons. How our brain perceive and interpret sound is very different from our instruments and computers.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,836
Likes
16,503
Location
Monument, CO
Yes, your job was on different aspect of audio and photons. How our brain perceive and interpret sound is very different from our instruments and computers.

Interesting, thank you. I'll bow to your and John's experience.
 

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,597
Likes
2,235
Maybe you can tell me what you think I think?

How could I possibly do that?

Anechoically flat means input equals output (again, along some solid angle). The room is the confounding factor, of course, ignoring near-field listening for the time being.

Imo there are two things a loudspeaker should get right: The direct sound, and the reflections. Getting the former right is rather straightforward; getting the latter right is a lot more complicated.

If direct sound were straightforward to get right, I would naively assume manufacturers would routinely get that right. Alas...

One cannot get all rooms (reflected sound) right, so I'm not sure what magic sauce you think you've found. I am curious though.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,525
Likes
37,058
Understood, of course, but we were talking in-room measurements, of the type often displayed here, and obsessively refined and adjusted, until a satisfactory result is achieved - which arithmetically could be achieved just as fast with any FR from the first element.
Yes, and there is the known relationship between in room and anechoic response. The in room should have a shallow downward tilt as frequency increases. This is the in room measure of an anechoicly flat speaker.

In room you also have issues below the Schroeder frequency, but getting that flat at the LP is in the proper direction. A bit above these frequencies you are hearing the balance of the speakers. If EQ can improve them you'll hear that too. Our hearing largely filters out early reflections helping us hear the speakers response with less involvement from the room.
 

escksu

Addicted to Fun and Learning
Joined
Jul 16, 2020
Messages
965
Likes
397
How does a person know if the speaker the reviewer reviewed "sounds" the same as the one the person is buying? Which means how do the manufacturer tell if the units they are making in the production line are the "same" (to within some tolerances, which are very rarely disclosed)?

I haven't heard any company uses listening as a method for QA/QC. (I am talking about pass/fail based on the judgements by some "qualified" QC person on the "sound quality", not if the unit makes sound.) Aren't all these all done by measurements?

If there are some magical qualities in the product that are not measurable, how does the manufacturer make sure they are in every unit they are selling, in the correct amounts?

Give you a clue... musical instruments.

Btw, alot more industries are human subjective based QC... Alcoholic drinks and tea for instance, nobody measures how good/bad a liquior/wine etc is (lab test alcohol level), so you still have someone doing taste and smell test before approving it.

Same for tea, blending of tea is done by an expert who taste them and determine how it should be blended.

Motorsports racing (esp. F1) is an even bigger one... It all depends on the race driver's feel, not measurements. Showing measurements doesnt work. The driver doesnt feel good means it doesnt work. Eg. Driver prefer AP brakes over brembo ones (or vice versa) because it feels better... It doesnt matter what the numbers say. you may even hear drivers say a new set of tyres dont feel right and get it replaced. At the end of the day, driver confidence matters more than any measurement.
 

escksu

Addicted to Fun and Learning
Joined
Jul 16, 2020
Messages
965
Likes
397
Interesting, thank you. I'll bow to your and John's experience.

Anything that deals with humans esp. psychological part isnt as straight forward as making measurements and getting numbers. There are a whole lot of human related aspects that are simply not measurable.

Things like feelings. Even the simple feeling of joy, sad, pain, love etc.. We dont have a way to measure them. its not like you can connect a machine and know if the person is indeed feeling sad or rating on the sadness. We can only depend on how the individual's description and sometimes expression. Telling lies... Our so-called lie detectors are simply measuring how a normal person usually react when telling lies (elevated heart beats, feeling anxious etc). But they are those who are able to tell such a convincing lie you thought its the truth.

Regarding psyshoacoustics, perhaps you may come across a person thinking the sound appears from a certain place. But the sound isnt from there. Its what the brain perceived.
 

escksu

Addicted to Fun and Learning
Joined
Jul 16, 2020
Messages
965
Likes
397
At the end of the day, both measurements and hearing tests are actually correct. Measurements are more accurate but not everything can be measured (at least not for now), humans are subjective and hearing etc may not be accurate. All have their pros and cons. Its up to you on what you prefer and believe. No idea whats all the argument about.....
 

Duke

Major Contributor
Audio Company
Forum Donor
Joined
Apr 22, 2016
Messages
1,523
Likes
3,745
Location
Princeton, Texas
Anechoically flat means input equals output (again, along some solid angle).


Agreed. Apologies if I came across as if I thought something different.

The room is the confounding factor, of course...


Loudspeaker/room interaction can be overall beneficial or overall detrimental. The more correct the reflections, the more beneficial the loudspeaker/room interaction is likely to be, in my opinion.

One cannot get all rooms (reflected sound) right, so I'm not sure what magic sauce you think you've found. I am curious though.


Floyd Toole has said that "frequency response is the most important parameter in any audio component. If it is wrong, nothing else matters." So for now, let's just look at frequency response, because THIS is what matters the most. In particular, let's pay attention to the influence the reflected sound has on a speaker's frequency response (we can come back to the effects of reflections on other things later).

Now in the context of loudspeakers I THINK Toole means the "perceived frequency response", rather than one PARTICULAR frequency response measurement (anechoic, in-room, whatever). The direct sound is not the perceived frequency response, nor is the reflected sound, nor is the measured steady-state in-room response (which includes both). That which is perceived to be "flat" may well include a flat direct sound plus non-flat reflected sound. The tonal balance we perceive is in effect a weighted average of the direct and reflected sound, with that weighting probably being tipped towards the first-arrival sound by an amount which varies with the characteristics of the reflections at the listening position.

We have a body of research which shows that a combination of flat direct sound and non-flat reflected sound which sums to a steady-state in-room frequency response curve along the lines of those shown in post #13 results in a subjectively flat, or at least subjectively preferred, perceived frequency response. (The "trained listeners" curve is arguably the most reliable one.)

So here is the big question: Is a combination of flat direct sound plus non-flat reflected sound the ONLY valid way to achieve our preferred perceived frequency response?

Maybe not. Let me describe one theoretical alternative approach:

Suppose we prioritize minimizing the spectral discrepancy between the direct and reflected sound, such that the direct sound's frequency response becomes MORE like the (non-flat) preferred steady-state in-room frequency response, and the reflected sound becomes MORE like the direct sound (closer to flat)? In other words, have the two approximately "meet in the middle". Might there be a subjective benefit to having the spectral balance of the reflections be virtually identical to that of the first-arrival sound? Well, we don't know - it hasn't been researched (though I can think of arguments why it might be). Is there an obvious reason why this approach would not be viable? If so let me know, because such is not apparent to me.

Let me know your thoughts thus far, if you don't mind. (And no, I'm not necessarily talking about an omnidirectional system, though that is one possible approach.)
 
Last edited:

JSmith

Master Contributor
Joined
Feb 8, 2021
Messages
5,155
Likes
13,220
Location
Algol Perseus
How does a person know if the speaker the reviewer reviewed "sounds" the same as the one the person is buying?
You won't... it's a guide. Like everything purchased in life, caveat emptor.



JSmith
 

jae

Major Contributor
Joined
Dec 2, 2019
Messages
1,208
Likes
1,508
Perfect and complete loudspeaker measurements perfectly and completely interpreted should (and imo would) be accurate predictors of loudspeaker preference. Imo we are not there yet.

My understanding is that spin-o-rama analysis is something like 87% accurate in its predictions of loudspeaker preference among those loudspeakers whose measurements were used to derive the predictive algorithms. That's very good but not yet perfect. (What's going on with the other 13%?)

Here are a few areas where loudspeaker measurements are arguably not as predictive as we might hope:

- The floor-bounce dip and subsequent peak usually stick out like sore thumbs in measurements which include them, but are seldom even noticed by the ear.

- A 5-inch woofer and a 15-inch woofer may have identical frequency response curves but will sound very different even at the same SPL. So we cannot equalize a minimonitor to sound like a big JBL.

- Reports of subwoofers subjectively improving the mids and highs (whether or not the mains are high-pass-filtered), and add-on supertweeters subjectively improving the bass, have no correlation with measurements.

- It is not generally apparent from measurements which horns have "horn coloration" and which ones do not.

- If we have two identical sounds, except that one lasts a little bit longer than the other, the longer one will be perceived as being louder, despite the measured SPL being identical.

- Measurements fail to predict imaging, much less spaciousness. Research by Wolfgang Klippel (cited by Floyd Toole) indicates that “the feeling of space” makes a 50% contribution to “naturalness” and a 70%(!) contribution to “pleasantness”. The virtual uselessness of measurements in this area is arguably a significant shortcoming if Klippel's findings are in the ballpark.

- Harman's subjectively-preferred steady-state in-room response curves, whether blind-generated by trained listeners or untrained listeners or both or based on extensive loudspeaker preference data, are all non-flat (and non-constant-slope) and counter-intuitive.

So imo we are not yet at the point where we can eyeball a few curves and make fully informed decisions about loudspeaker preference.

Despite timbral perception in multidimensional space not being completely understood, we do know enough that some of the things you mentioned can well be quantified by measurements pertaining to impulse/acoustic envelope, dispersion/directivity, and other measurable quantities pertaining to amplitude/frequency/phase dynamics. Just because they are not typically done, does not mean we can't explain and reproduce some of these phenomena right now if so desired. There has to be some genuine motivation or incentive to do so, however. Most are still struggling with just reliably reproducing a desired spectrum/frequency response and understanding how it affects their sound.
 
OP
G

gn77b

Member
Joined
Oct 30, 2020
Messages
77
Likes
68
1. The measurements made currently do not span the psychoacoustic space.

2. Factors can act in combination but this is not accounted for.
Sorry but that's just typical audiophile hand waving, you sound just like a politician. You didn't really say anything that wasn't explicitly said or implied by my original post.
 
Top Bottom