• 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!

Research Project: Infinity IL10 Speaker Review & Measurements

pozz

Слава Україні
Forum Donor
Editor
Joined
May 21, 2019
Messages
4,036
Likes
6,827
Trouble is, by the looks of it, it may not be very different from the M16 in that regard. Perhaps in combination with other factors it could come off that way. Here's how the IL10 looks like vs the M16 with Amir's measurements (ER curve corrected with the proper formula):
index.php
Shows just how much eyeballing the graphs independently misses.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,986
Location
Riverview FL
On Distortion:


I find my little JBLs to become unappealing at higher levels. They're fine for casual listening, especially if not being dmanding of the imaging.

It doesn't have to be too too high a level before they create that "grunge" sound.


Two speakers: JBL LSR 308 and MartinLogan reQuest

The frequency response, after correction, as measured here, at the listening position (10 feet), of each, is very similar - target was "flat".

That leaves dispersion and distortion as suspects for the displeasure.

Dispersion differences affect the 'stereo", but don't change distortion.

So, distortion seems (to me) to be the fault noticed at higher SPL with the little JBLs.




Simplistic measure:

Speakers in the same room, JBL adjacent to but outboard of the ML.

82dB is not a particularly loud level, less than a watt of power applied.

JBL cone


index.php



ML electrostat


index.php



Swept sine THD, dB relative at a higher level:

1593116419482.png


And THD % relative to the fundamental

1593116587492.png



*the JBLs are my daily drivers for most video and casual music when I'm inattentive. When I really want to hear music, I power up (literally) the ML. They've never sounded "strained". 110dB unweighted "flat FR" peaks, no problem. But, they're big speakers, and I have plenty of amplification to goose them.
 
Last edited:

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
I don't read anything specific to distortion here.

Read again the 1st post and you'll find this:

"Combine that with the fact that unless one is trained, hearing distortion is difficult and you or at least I arrive at the fact that distortion matters once you take care of tonality. "

All this discussion about Amir's hearing abilities is beside the point. Is your point to establish that his judgement is unsound and inconsistent? What's the use in that? Address the data and find more testing, not the person. By the way ad hominem doesn't mean "insults", it is an invalidation of an argument by calling attention to the person making it.

My point is that any sighted listening contains bias and trained listeners are not immune to that. I have backed up that claim with a link from Olive's blog containing his research. As my point is not related to Amir it cannot be "ad hominem" - I stated it clearly in this post, so you may consider first reading someones post before accusing him of ad hominem attack.

Let me put it simple for you: seeing the distortion measurement first and than claiming you are hearing it in a sighted listening test and that it is not related to sighted bias because you are a trained listener is simply false, as explained in Olive's article.
 
Last edited:

Vuki

Senior Member
Joined
Apr 8, 2018
Messages
341
Likes
392
Location
Zagreb, Croatia
IME- grunginess and lack of clarity heard by @amirm comes from untamed midbass driver resonance at 4.7kHz. I made few loudspeakers with midbasses that had such resonance (metal/ceramic membrane) and one can easily hear it if not tamed by the crossover.
 

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,404
Likes
5,296
Location
Somerville, MA
Also my two cents on the importance of the spinorama. I don't th
I'm curious if it's possible to create a number of very similar bookshelves that are largely the same in as many areas possible but with specific dips in DI, for ease of listening comparison, with slightly different baffle shapes, for example.

So if I am reading this debate here correctly, it seems like you want to isolate the effect of DI on sound quality perception, specifically trying detach it from other factors such as distortion. Obviously you can equalize a speaker to have flat FR on axis, which eliminates that factor. To eliminate the influence of distortion, you have two options. One, use the same drivers, or two, use drivers which have negligible distortion.

Being an idealist, let's try to pursue option #1. Unfortunately, the biggest contributors DI are piston size, so the range of different DI will be small. Yes, you could make a wide baffle and narrow baffle speaker, you could make a poorly integrated or well integrated crossover, but you start running into problems when you try to manipulate the drivers into having different dispersion characteristics.

All things being equal, a speaker radiating in a wider pattern will be putting more sound power into the room. So, if you have a narrow dispersing driver and a wide one, both the same amplitude on axis, the narrow one will be having an easier time doing it. In other words, distortion goes down with narrow dispersion and goes up with wide dispersion. At the extremes we can illustrate this principle easily - dipole speakers need to move like crazy to get mid-bass, and tweeters in waveguides can be tiny little domes and still play loud without difficulty (see the 19mm unit in the buchardt.)

So, using the same drivers, you can change it a bit. Smooth vs lumpy, maybe massage the upper midrange a few db, but if you want something very different you need to use different drivers.

If you want to isolate the importance of distortion I might suggest a simpler test - go outside, somewhere quiet and free from lateral reflections, and play the same speaker at different distances, but keeping volume at listening position the same. Pretty easy to do a good test.
 

GelbeMusik

Senior Member
Joined
Apr 19, 2020
Messages
445
Likes
290
I'm talking Klippel measurements specifically. That level of precision is necessary.

We all talk Klippel / Toole / Olive, aren't we?

I get tired of discussing again and again any insights of the three gentlemen down to the smallest detail. Sometimes, please do not be offended, it reads like a biblical interpretation. Obviously, some of the participants have not had much contact with science. I live in it, trapped. Poor me. I need some air.
 

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,404
Likes
5,296
Location
Somerville, MA
Daily reminder that harmonic distortion is an engineering conceit and not a perceptual metric. HD numbers for a woofer can sort of indicate the comparitive ease with which it will operate outside of its comfort zone, but if we're talking midrange on up it is not the correct thing to be looking at. It has been determined fairly convincingly by the Gedlee research that 'the thing we're talking about' when we discuss "distortion" in sound quality discussions is not the same thing that we are measuring when we look at harmonic or IM distortion charts. We can see the limitations of these graphs every time audiophiles say they prefer second order, or even order, but not odd order! distortion. What kind of IM distortion is most euphonic I have no idea, but if we're talking about the subtle grunginess in the mids and treble, harmonic distortion may point to the problem but it is an unreliable and useless way to compare and measure.
 

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,111
Likes
8,443
Location
NYC
Shows just how much eyeballing the graphs independently misses.
Yeah. In this case I'd been looking at Harman's spins since they have the correct ER calculation, In harman's spins it's evident the IL10 has a scoop while the M16 is more of a flat line. Of course, there's always sample variation too...


This is early in analysis but i received a Neumann KH80 and measured it. The results now don't show that slight dip that it did in the first two tries. The running hypothesis between Neumann and I are that temperature makes a difference. I measured M16 and KH80 during winter and it seems that may have impacted bass.

I need to test more but there just isn't enough time to keep up with huge amount of gear that has arrived let alone go back and redo tests.
Now that is quite interesting.
 

617

Major Contributor
Forum Donor
Joined
Mar 18, 2019
Messages
2,404
Likes
5,296
Location
Somerville, MA
We all talk Klippel / Toole / Olive, aren't we?

I get tired of discussing again and again any insights of the three gentlemen down to the smallest detail. Sometimes, please do not be offended, it reads like a biblical interpretation. Obviously, some of the participants have not had much contact with science. I live in it, trapped. Poor me. I need some air.

Or how about when Amirm says that he's professionally trained to hear minor traces of distortion like he's some kind of audiophile Jason Bourne.
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
It has been determined fairly convincingly by the Gedlee research that 'the thing we're talking about' when we discuss "distortion" in sound quality discussions is not the same thing that we are measuring when we look at harmonic or IM distortion charts.

Some quotes from Geddes/Lee study about distortion perception:

"Now consider the loudspeaker example. Unless it has some severe design or manufacturing problems, it will mostly have lower orders of nonlinearity and the distortion will typically rise with level. Based on our principles, we should expect this type of distortion to be fairly benign, almost inaudible, and this is in fact what we find to be true (for comparable levels of THD for the loudspeaker and the amplifier).

The root cause of distortion is the nonlinearity of the system and the correct way to discuss nonlinearity is with the orders of its nonlinear transfer function. When one views the distortion problem in this way, signal based distortion metrics (THD, IMD, etc.) become irrelevant. "
 

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,111
Likes
8,443
Location
NYC
Daily reminder that harmonic distortion is an engineering conceit and not a perceptual metric. HD numbers for a woofer can sort of indicate the comparitive ease with which it will operate outside of its comfort zone, but if we're talking midrange on up it is not the correct thing to be looking at. It has been determined fairly convincingly by the Gedlee research that 'the thing we're talking about' when we discuss "distortion" in sound quality discussions is not the same thing that we are measuring when we look at harmonic or IM distortion charts. We can see the limitations of these graphs every time audiophiles say they prefer second order, or even order, but not odd order! distortion. What kind of IM distortion is most euphonic I have no idea, but if we're talking about the subtle grunginess in the mids and treble, harmonic distortion may point to the problem but it is an unreliable and useless way to compare and measure.

And going back to the paper being discussed in this thread, there's a whole section about how what Listeners described as distortion were mostly just tied to their preferred speakers rather than actually hearing distortion. The correlation between listeners' rated distortion and their preference was 0.86 lol. I don't know if any were specifically trained to hear distortions like Amir, I know that's not part of Harman's usual training, but it goes to show that what people often describe as distortion is not actually distortion.

Snag_1f3f7246.png


That said, there were significant differences among individual listeners.
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
And going back to the paper being discussed in this thread, there's a whole section about how what Listeners described as distortion were mostly just tied to their preferred speakers rather than actually hearing distortion. The correlation between listeners' rated distortion and their preference was 0.86 lol. I don't know if any were specifically trained to hear distortions like Amir, I know that's not part of Harman's usual training, but it goes to show that what people often describe as distortion is not actually distortion.

View attachment 70710

That said, there were significant differences among individual listeners.

Is this the same speaker with increase distortion or does each Lx refer to a different speaker?
If the latter that chart is meaningless.
 

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
2,111
Likes
8,443
Location
NYC
Is this the same speaker with increase distortion or does each Lx refer to a different speaker?
If the latter that chart is meaningless.

The latter, but that's kind of the point. Listeners were largely unable to differentiate their perception of distortion from their overall perception of preference. Measured distortion did have a weak negative correlation with preference, but not enough to be statistically significant. It just so happened that on average, the speakers with lower distortion ratings also tended to have more linear frequency response.

1593119281414.png


Anyway, not trying to say distortion doesn't matter at all or that Amir can't hear them. Just addressing the larger point of many audio enthusiasts thinking they can readily hear distortion.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,538
Location
Seattle Area
Daily reminder that harmonic distortion is an engineering conceit and not a perceptual metric.
The summed value (THD) is. Its harmonic structure however can be fully psychoacoustically analyzed and audibility determined. The measurements I show have the harmonic structure in different colors and lines. This is how the earlier audibility analysis was done.

So yes, it is not automated like "THD> X is audible" but it is supremely valuable in determine non-linearities and their audibility once the spectrum is analyzed. Indeed it is the gold standard in research to determine absolute transparency.
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
From Geddes/Lee AES convention paper "Auditory Perception of Nonlinear Distortion":

"A new approach to the perception of distortion was recently proposed by Geddes (2002). Psychoacoustical data were measured, correlation and regression analysis were applied to examine the relationship and predictive value of this new metric to the subjective assessment of sound quality of nonlinear distortion. Furthermore, conventional metrics such as total harmonic distortion (TDH) and intermodulation distortion (IMD) were also compared. Thirtyfour listeners participated in a listening task, rating twenty-one stimuli using a 7-point scale. No significant relationships were observed when comparing the subjective ratings with TDH and IMD metrics. Significant correlation (r=0.95, p<.001) was observed between the subjective ratings and the new proposed GedLee (Gm)metric. "
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,538
Location
Seattle Area
Yep, I know, this one makes it really hard for you to talk your way out of it. I'm sure you are aware that reputation is easilly lost when you preach one thing in the morning and another in the evening so I'll just leave you to it.
It is the difference between theory, reading stuff online, and practicing and understanding it.

There is no black and white here. Even the best double blind controlled test has limitations, degrees of error, etc.

As I explained to you, it is indeed possible to show sources of errors even with trained listeners. What you are not appreciating is that the industry deals with degree of error in an effective way to engineer products.

Do you really think someone at harman goes and performs months of blind testing every time they want to make a small change to the speaker in its design process?

I have managed the development of audio technologies that absolutely rely on controlled testing for their results: lossy audio compression. I can tell you 100% that vast majority of development is performed using trained listeners in sighted environment. A trained listener here is doing a job, he gets paid to do it correctly.

You think the trained listener sampling bear in each run is ignored and double blind tests conducted every day instead?

Sean told me they abandoned efforts to develop the tool to determine preference testing for Harman speaker development. If you think the results from research are perfect because they were double blind, why did they do this?

So no, it is different when a trained professional evaluates products in his field of expertise than average joe. You cannot create equivalence about them. Your doctor's opinion about what is wrong with you is much more accurate than you doing the same analysis by googling.

Learn to judge value of data points. Don't make blind conclusions and assertions.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,538
Location
Seattle Area
From Geddes/Lee AES convention paper "Auditory Perception of Nonlinear Distortion":

"A new approach to the perception of distortion was recently proposed by Geddes (2002). Psychoacoustical data were measured, correlation and regression analysis were applied to examine the relationship and predictive value of this new metric to the subjective assessment of sound quality of nonlinear distortion. Furthermore, conventional metrics such as total harmonic distortion (TDH) and intermodulation distortion (IMD) were also compared. Thirtyfour listeners participated in a listening task, rating twenty-one stimuli using a 7-point scale. No significant relationships were observed when comparing the subjective ratings with TDH and IMD metrics. Significant correlation (r=0.95, p<.001) was observed between the subjective ratings and the new proposed GedLee (Gm)metric. "
I just explained what this means and you go and quote it again? It is clear that you are not reading the answers given to you. Nor do you have any appreciation of what you are quoting.
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
It is the difference between theory, reading stuff online, and practicing and understanding it.

There is no black and white here. Even the best double blind controlled test has limitations, degrees of error, etc.

As I explained to you, it is indeed possible to show sources of errors even with trained listeners. What you are not appreciating is that the industry deals with degree of error in an effective way to engineer products.

Do you really think someone had harman goes and performs months of blind testing every time they want to make a small change to the speaker in its design process?

I have managed the development of audio technologies that absolutely rely on controlled testing for their results: lossy audio compression. I can tell you 100% that vast majority of development is performed using trained listeners in sighted environment. A trained listener here is doing a job, he gets paid to do it correctly.

You think the trained listener sampling bear in each run is ignored and double blind tests conducted every day instead?

Sean told me they abandoned efforts to develop the tool to determine preference testing for Harman speaker development. If you think the results from research are perfect because they were double blind, why did they do this?

So no, it is different when a trained professional evaluates products in his field of expertise than average joe. You cannot create equivalence about them. Your doctor's opinion about what is wrong with you is much more accurate than you doing the same analysis by googling.

Learn to judge value of data points. Don't make blind conclusions and assertions.

Don't discuss it with me, discuss it with Dr. Olive:

A Blind Versus Sighted Loudspeaker Experiment

This question was tested in 1994, shortly after I joined Harman International as Manager of Subjective Evaluation [1]. My mission was to introduce formalized, double-blind product testing at Harman. To my surprise, this mandate met rather strong opposition from some of the more entrenched marketing, sales and engineering staff who felt that, as trained audio professionals, they were immune from the influence of sighted biases.

In summary, the sighted and blind loudspeaker listening tests in this study produced significantly different sound quality ratings. The psychological biases in the sighted tests were sufficiently strong that listeners were largely unresponsive to real changes in sound quality caused by acoustical interactions between the loudspeaker, its position in the room, and the program material. In other words, if you want to obtain an accurate and reliable measure of how the audio product truly sounds, the listening test must be done blind.
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
I just explained what this means and you go and quote it again? It is clear that you are not reading the answers given to you. Nor do you have any appreciation of what you are quoting.

Don't discuss it with me, discuss it with Dr. Geddes.

"A new approach to the perception of distortion was recently proposed by Geddes (2002). Psychoacoustical data were measured, correlation and regression analysis were applied to examine the relationship and predictive value of this new metric to the subjective assessment of sound quality of nonlinear distortion. Furthermore, conventional metrics such as total harmonic distortion (TDH) and intermodulation distortion (IMD) were also compared. Thirtyfour listeners participated in a listening task, rating twenty-one stimuli using a 7-point scale. No significant relationships were observed when comparing the subjective ratings with TDH and IMD metrics. Significant correlation (r=0.95, p<.001) was observed between the subjective ratings and the new proposed GedLee (Gm)metric. "
 

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,785
I just explained what this means and you go and quote it again? It is clear that you are not reading the answers given to you. Nor do you have any appreciation of what you are quoting.

@pozz, would this be ad honimem?

In my post I clearly stated that my opinion is not related to Amir in person:
"Btw, don't take it personal, I wouldn't trust Olive's sighted listening test done in your room after peeking at measurements. I simply don't believe in sighted test done by a single person. I treat it as personal opinion, not measured fact. "

But here Amir's post seems to be addressed to me personnaly and not to my opinion, which perfectly fits into "ad hominem" definiton, so do you dare to ask him "if that is really necessary"? ;)
 
Last edited:
Top Bottom