Its possible that the dominance of second order distortion helps reduce fatigue. It acts a little bit like a blanket over the sound.
I believe car manufacturers use a lot of cheap parts but do research on the internal cabin and things like positioning and dispersion to get most from their investment. This supports the idea that things like room and dispersion matter the most.
If I was put on the spot I'd say linear distortion matters most, then a close second is the speaker environment interaction (including dispersion and reflections), which is a much more complicated variable.
Nonlinear distortion is more for the refined pallet. A lot of recordings and sounds have high levels of distortion inherently so it's easier to adjust to.
As you move closer to source (near field listening) the linear and nonlinear matter more and the room less, which seems obvious. If you have a difficult room it might be easiest to build low distortion easy listening speakers or use headphones. I find a larger baffle also helps limit room interaction. Zaph's ZRT might work well.
Nonlinear distortion is more for the refined pallet. A lot of recordings and sounds have high levels of distortion inherently so it's easier to adjust to.
As you move closer to source (near field listening) the linear and nonlinear matter more and the room less, which seems obvious. If you have a difficult room it might be easiest to build low distortion easy listening speakers or use headphones. I find a larger baffle also helps limit room interaction. Zaph's ZRT might work well.
Is this a "known thing," do drivers at low output levels generate substantially less distortion than at higher, "moderate" (but well within linearity, xmax, power ratings and such) levels?
one interesting observation i've made over my tenure in the live audio is that percentage of distortion trumps volume every time!
2 watts of heavily distorted guitar can drive people out of room (or have them wailing "it's to damn loud!") but equivalent volume of pop or country is barely noticed.
as one who has lost sleep over understanding the nature and perception of distortion both as a problem, and yes, a tool/effect.
i've become pragmatic and accept that in some circumstances distortion can be a problem in others it can be a benefit.
2 watts of heavily distorted guitar can drive people out of room (or have them wailing "it's to damn loud!") but equivalent volume of pop or country is barely noticed.
as one who has lost sleep over understanding the nature and perception of distortion both as a problem, and yes, a tool/effect.
i've become pragmatic and accept that in some circumstances distortion can be a problem in others it can be a benefit.
This sounds like another effect rather than non-linear distortion. It can also happen with compression tweeters.Zarathu said:
On the other hand the derivative of variations in the transfer function versus level in an amp can lead to effects in a system.
Same experience here.
A small PA driven hard into distortion will always be perceived as much louder than a large one that is just loafing when both produce the same SPL in the same room. This is a well known effect in live audio.
The other thing I noticed is if I go to a gig and the band cheapened out settled for a small, distorting PA after the gig my ears ring for some hours.
Same venue, large quality PA and the same SPL levels result in much less or no ear ringing afterwards.
From Floyd Toole's book--1st and 2nd editions, pg. 382--talking about a JBL horn-loaded loudspeaker under development (probably the K2 M9500):
I actually found Toole's comment to be hilarious, considering how much time and effort he spent talking about direct radiating loudspeakers.
Chris
Note that he removed these passages on testing horn-loaded loudspeakers from his latest 3rd edition. Paul Klipsch mentions modulation distortion as the source of this type of distortion--AM and FM distortion--and is very apparent to the ears at levels well above 80 dB(A).
I actually found Toole's comment to be hilarious, considering how much time and effort he spent talking about direct radiating loudspeakers.
Chris
Last edited:
It's pretty common knowledge that heavy distortion sounds loud - millions of guitar players know that - but that isn't what Geddes is taking about. At least as far as I've read his research results. Below a certain level THD just does not make an audible difference in the perception of a loudspeaker.
Earl Geddes will correct me if that's not accurate.
Earl Geddes will correct me if that's not accurate.
I think Earl can defend himself here...and I do think that was what Earl was "talking about".
The notion that higher SPL nonlinear distortion (particularly modulation) isn't relevant to hi-fi loudspeakers in the home and in other applications, such as cinema and PA is only true to those that want to believe that's true, in my experience.
My ears have testified to Klipsch's and Keith Howard's findings (among others), however.
Chris
The notion that higher SPL nonlinear distortion (particularly modulation) isn't relevant to hi-fi loudspeakers in the home and in other applications, such as cinema and PA is only true to those that want to believe that's true, in my experience.
My ears have testified to Klipsch's and Keith Howard's findings (among others), however.
Chris
Let me clear this up.
The statement that I was responding to said, as I recall, that loudness was governed by how much distortion there was. It is simply absurd to think that distortion is the most significant attribute of loudness. (Lidia did her PhD in loudness, so I have some background here.) Loudness is a subjective term and many things can contribute, but distortion is not the principle one - that would be SPL and then spectrum. AT normal levels nonlinear distortion does not even enter the picture.
I did say that distortion would be a factor, where I meant, once the distortion was significant enough to be audible. As always, I am mostly referring to home usage where maximum SPL is NEVER a main attribute. If you are listening to your home system at a point where the distortion is audible then you simply need a more powerful system.
In live venues where distortion is commonplace there will be an incremental effect to loudness from the distortion in two ways (I said this in my initial response.) First is that the spectrum moves upwards in frequency, all distortion products go higher in frequency, which makes it sound louder. There is more energy in the signal, which makes it sound louder, and its more annoying, which makes it sound louder. These factors all grow very rapidly once a system goes into overload - as we all have witnessed.
But, at normal home listening levels, for a system like mine distortion would be completely irrelevant to the perceived loudness.
Now there is also a confounding aspect that one cannot ignore here and that is the fact that diffraction become more and more audible with SPL. There are no studies that I know of to differential if the incremental loudness with level is due to nonlinear distortion like THD or diffraction. It could be either or both, it would be a tough test to discriminate.
The statement that I was responding to said, as I recall, that loudness was governed by how much distortion there was. It is simply absurd to think that distortion is the most significant attribute of loudness. (Lidia did her PhD in loudness, so I have some background here.) Loudness is a subjective term and many things can contribute, but distortion is not the principle one - that would be SPL and then spectrum. AT normal levels nonlinear distortion does not even enter the picture.
I did say that distortion would be a factor, where I meant, once the distortion was significant enough to be audible. As always, I am mostly referring to home usage where maximum SPL is NEVER a main attribute. If you are listening to your home system at a point where the distortion is audible then you simply need a more powerful system.
In live venues where distortion is commonplace there will be an incremental effect to loudness from the distortion in two ways (I said this in my initial response.) First is that the spectrum moves upwards in frequency, all distortion products go higher in frequency, which makes it sound louder. There is more energy in the signal, which makes it sound louder, and its more annoying, which makes it sound louder. These factors all grow very rapidly once a system goes into overload - as we all have witnessed.
But, at normal home listening levels, for a system like mine distortion would be completely irrelevant to the perceived loudness.
Now there is also a confounding aspect that one cannot ignore here and that is the fact that diffraction become more and more audible with SPL. There are no studies that I know of to differential if the incremental loudness with level is due to nonlinear distortion like THD or diffraction. It could be either or both, it would be a tough test to discriminate.
Ahem.. you might want to check on that before making such a statement.
It even has a title with "Subjective Evaluation" in it.
As for Zaph, his argument is based on his own perceptual results and those of others (..though under a sighted condition) which owes more to a perceived effect generated from longer-term listening. He argues long-term listening learned behavior and a resulting less fatiguing loudspeaker with lower non-linear distortion.
Griesinger's "Perception of Mid and High Frequency.." power point is interesting (starting on slide 39).
It's somewhat odd, but I'd side with Zaph here.. though at the same time I think he places to much emphasis on it - presumably for lack of concentrating on other areas of loudspeaker distortion other than the most apparent design elements of on-axis or listening-axis linearity. (..sort of a: "once you get a basic design down that's linear you should be focusing heavily on non-linear distortion reduction because that's all that's left.") I personally do NOT subscribe to that thought process - there are other effects in the linear realm that are not at all well studied that have a far more substantive effect than modest levels of non-linear distortion vs. low levels of non-linear distortion. (..and I have selected drivers with more non-linear distortion vs. lower distortion drivers for *other* reasons beyond basic linearity.)
Perhaps he dabbled in it, but one paper with the word "subjective" in the title does not an expert make. I am not aware of any journal publications by him on the topic. But the point is that I would not put Klippel in the same ballpark of that subject matter as Toole/Olive. Objective measurements is his expertise.
In lieu of reading his book, I just watched Toole's presentation on youtube.
YouTube
It seems what he's advocating is flat frequency response, aka low linear distortion, at all angles. I don't think he puts much emphasis on nonlinear but he doesn't address it directly.
There are some things that bother me about this. Why on earth would harmon which owns JBL fund all this research only to go out and make it all public and then just happen to have the "magic formula" already mastered in their overpriced sizzle and boom boxes. Kind of seems like they got the result they wanted.
He also manages to knock competitors like ML and KEF and then sort of poo poo the smaller companies that don't do his kind of mega-research.
When you put speakers in a large empty test room then the importance of off axis response goes way up, because the sounds going to bounce around more. I'm just not all that impressed with the experiments, not saying its a conspiracy or anything.
He does mention reducing the resonance, ringing and box noise all of which I agree with. My own experience with seos horn speakers is that they are incredible but not great nearfield and really excel at distance in reverberant spaces. The speakers I mentioned before (first build array wwtww) have seos horns for the treble. They are as wonderful from a distance as they are terrible close up.
Basically my hypothesis is that off axis linear distortion matters more as you move further away and in livelier spaces, so a CD horn like JBL sells works well in a theatre type space. Non linear distortion and things like driver integration matter more nearfield.
Also I should point out that Zaph also does off axis measurements and seems to value a smooth off axis response. I think the conventional wisdom is high dispersion and smooth transition to tweeter but the extreme off axis (over 45 degrees) isn't hugely important.
YouTube
It seems what he's advocating is flat frequency response, aka low linear distortion, at all angles. I don't think he puts much emphasis on nonlinear but he doesn't address it directly.
There are some things that bother me about this. Why on earth would harmon which owns JBL fund all this research only to go out and make it all public and then just happen to have the "magic formula" already mastered in their overpriced sizzle and boom boxes. Kind of seems like they got the result they wanted.
He also manages to knock competitors like ML and KEF and then sort of poo poo the smaller companies that don't do his kind of mega-research.
When you put speakers in a large empty test room then the importance of off axis response goes way up, because the sounds going to bounce around more. I'm just not all that impressed with the experiments, not saying its a conspiracy or anything.
He does mention reducing the resonance, ringing and box noise all of which I agree with. My own experience with seos horn speakers is that they are incredible but not great nearfield and really excel at distance in reverberant spaces. The speakers I mentioned before (first build array wwtww) have seos horns for the treble. They are as wonderful from a distance as they are terrible close up.
Basically my hypothesis is that off axis linear distortion matters more as you move further away and in livelier spaces, so a CD horn like JBL sells works well in a theatre type space. Non linear distortion and things like driver integration matter more nearfield.
Also I should point out that Zaph also does off axis measurements and seems to value a smooth off axis response. I think the conventional wisdom is high dispersion and smooth transition to tweeter but the extreme off axis (over 45 degrees) isn't hugely important.
Last edited:
Actually the point of the paper wasn't its subjective results, rather the focus was on "locking-down" the variables on the hardware end through auralization (with greater flexibility in testing that subjective response down to specific sources of distortion within drivers).
-basically a better way of conducting a subjective test on non-linear distortion.
Though note that the results were not terribly supportive of Zaph's argument. Nor was Griesinger's for that matter - and Griesinger specifically looked at long-term listening.
Talking of Klippel: On his website one can listen to various kinds of nonlinear driver disortion modes and check one's personal perception level for that particular kind of distortion.
You will have to use earphones in a quiet space.
Listening Test
As soon as your personal level has been determined you will see a histogram where you can compare your ability (or maybe tolerance to distortion) with others.
I repeatedly managed to be on the right half of the bell curve. But I must also say that even the levels of distortion that I could easily perceive were not always disturbing to me.
Regards
Charles
You will have to use earphones in a quiet space.
Listening Test
As soon as your personal level has been determined you will see a histogram where you can compare your ability (or maybe tolerance to distortion) with others.
I repeatedly managed to be on the right half of the bell curve. But I must also say that even the levels of distortion that I could easily perceive were not always disturbing to me.
Regards
Charles
I didn't feel like installing silverlight. From the experiments I've read our tolerance for non linear distortion goes up at frequency extremes and is much higher in actual music than tones.
Its possible that people can't A/B nonlinear distortion, but can over time appreciate less of it, similar to Zaph' ideas.
Its possible that people can't A/B nonlinear distortion, but can over time appreciate less of it, similar to Zaph' ideas.
Last edited: