Subwoofers: a need for lower distortion?

[RCAguy]

Gentlemen, I've enjoyed our conversation. However while citing a mgfr website and Wikipedia, none of it refers to the paper I introduced in the OP. No one has read the paper.

I'm not quite convinced that you've got my argument right as intended. Last chance?

You measure a speaker and then calculate the subjectively perceived strength of the harmonic content. Like you did above as 3% technical equals 25% subjectively @80dB at-the-ear level but subjectively 10% @90dB at-the-ear level.

Much of data. Now, what are you going to do with that?

To what subjective impression would you relate it? You're decidedly after subjective impressions, aren't you?

Well have you gotten my drift? I've agreed several times that subjective listening test - done properly (I've mentioned how) - is absolutely needed. And my paper is a first step: stating a hypothesis. I haven't proven it, but how would you know, you haven't read it. I get you loud & clear. AFAIK the only missing link is who's going to carry it through?

 

[fineMen]

No one has read the paper.

Sorry.

 

[RCAguy]

I agree, I just tried quickly drawing a fundamental at 30Hz with 90dB SPL and its harmonic at 60Hz with 60 dB SPL (i.e. roughly 3% THD) over the Fletcher-Munson equal loudness contour overlay:
View attachment 308530
We can see that 90 dB SPL at 30 Hz would be perceived equally loud as 80 dB SPL at 1kHz, while its 60Hz harmonic at 60 dB SPL would be perceived equally loud as ~25dB SPL at 1kHz - which gives a normalized difference of 55dB or approx. 0.2% normalized distortion.
But again, even such analysis disregards masking effects that happen when you introduce a second tone. To my knowledge equal loudness contours were derived using a single tone stimulus.
This is why I'd expect that controlled listening tests would show sensitivity to distortion at low frequencies is even lower than predicted by this kind of Fletcher-Munson normalization alone.

DOMINIKZ - Using ISO 226:2003 - 70yr newer data than F-M - Fig.2 (reproduced below) of the cited paper "Subwoofer Camp 2" shows that a pure 30Hz fundamental at 90SPL "sounds like" 40phons, while a 3% 2nd harmonic of 60Hz at 60SPL is heard as 20phons, a 20dB difference equivalent to 10% perceived distortion. And this "inflated" perception of LF distortion is greater still when other harmonics comprising THD are included, as in the paper. E.g. an often equal if not greater 3rd harmonic distortion product of 90Hz sounds like 30phons, 10dB below the fundamental, perceived as 30% distortion, etc. The paper is new analysis that refutes conventional wisdom on the perception of LF distortion.

Fig.2 of the cited paper "Subwoofer Camp 2"

We appreciate 'FineMen' and others above checking us on this.

We agree with 'FineMen's' points about conducting listening tests as an ultimate determination of the paper's hypothesis. And it pertains to all woofers reproducing <500Hz, not just SW. Such blind testing is beyond the scope of this hypothesis paper. But someone should do so - perhaps an academic with more resources than we have?

We also agree about the mitigating effect of masking. This is especially true with popular music that 1) to be novel contains copious intended harmonics; and 2) that is typically electric instruments for which listeners have little if any reference for their original undistorted sound. How fuzz-toned was it this time? However, masking has less effect when listening to acoustic music (classical, jazz, etc) for which most listeners have an accurate remembered reference, having attended unamplified concerts, or themselves playing acoustic instruments. For this once majority of listeners of reproduced music, low distortion means instruments are not artificially "brightened" by added distortion harmonics.

This reply is not intended to stymie further discussion - please weigh in. We all have a lot to learn!

PS - DominikZ - in my reposting this reply to you, deleting it in its original position also deleted your response - my apologies.

 

[dominikz]

DOMINIKZ - Using ISO 226:2003 - 70yr newer data than F-M - Fig.2 (reproduced below) of the cited paper "Subwoofer Camp 2" shows that a pure 30Hz fundamental at 90SPL "sounds like" 40phons, while a 3% 2nd harmonic of 60Hz at 60SPL is heard as 20phons, a 20dB difference equivalent to 10% perceived distortion. And this "inflated" perception of LF distortion is greater still when other harmonics comprising THD are included, as in the paper. E.g. an often equal if not greater 3rd harmonic distortion product of 90Hz sounds like 30phons, 10dB below the fundamental, perceived as 30% distortion, etc. The paper is new analysis that refutes conventional wisdom on the perception of LF distortion.

Fig.2 of the cited paper "Subwoofer Camp 2"

We appreciate 'FineMen' and others above checking us on this, however discussion might be more productive reading the paper first. (Limited time free access for ASR members by emailing me at [email protected].)

We agree with 'FineMen's' points about conducting listening tests as an ultimate determination of the paper's hypothesis. And it pertains to all woofers reproducing <500Hz, not just SW. Such blind testing is beyond the scope of this hypothesis paper. But someone should do so - perhaps an academic with more resources than we have?

We also agree about the mitigating effect of masking. This is especially true with popular music that 1) to be novel contains copious intended harmonics; and 2) that is typically electric instruments for which listeners have little if any reference for their original undistorted sound. How fuzz-toned was it this time? However, masking has less effect when listening to acoustic music (classical, jazz, etc) for which most listeners have an accurate remembered reference, having attended unamplified concerts, or themselves playing acoustic instruments. For this once majority of listeners of reproduced music, low distortion means instruments are not artificially "brightened" by added distortion harmonics.

This reply is not intended to stymie further discussion - please weigh in. We all have a lot to learn!

PS - DominikZ - in my reposting this reply to you, deleting it in its original position also deleted your response - my apologies.

Thanks for a thoughtful response!
As I said originally, while I found the approach interesting (and definitely applaud the effort to formalize it into a paper), I still have two main reservations:

1) The "sounds like" wording might be misleading to casual readers who might not catch that the hypothesis is not backed by formal listening tests

2) The equal-loudness contours (whether the older FM or the newer ISO curves) might not be the best target for normalization in this case since they do not take into account auditory masking, which neccessarily comes into play when two or more tones play simultaneously (like a fundamental and its harmonic). It would IMHO be interesting to try and normalize THD to some of the masking curves reported in earlier studies. But even so, formal listening tests would be needed to see how figures obtained in that way relate to audibility.

As it is I unfortunately don't find the hypothesis of inflated perceived LF distortion convincing enough to invest more time looking into it. But I'd be happy to change my mind should enough evidence come to light!

 

[fineMen]

And this "inflated" perception of LF distortion ,,,

So, what exactly is 'perception'? If the distortion component was singled out, in that the base tone (that what undergoes the distortion) is taken away but the distortion component is kept, then your musings are correct, but only tell what the F/M curves are about.

If you insist that the base tone is kept and heard with the distortion component together, the prerequisites / conditions for the F/M curves are no longer met. The F/M curves are for single tones, not for two or even more combined.

What chimes in with the combination of tones is the masking effect (s/ above many times). So regarding distortion you are in an entirely different regime. This mode as to say was also investigated by scientists in the field. We all know the results, which, to our all relief, indicate that the bass doesn't ask for that much of a concern regarding HD.

Hand your paper in to the Audio Engineering Society, really, they are keen on new stuff. Then you'll see if they give you an answer.

 

[rynberg]

wow, this is a very dramatic logical error. 30dB down at 1000Hz will psychologically roughly mean 60dB down at 25Hz. So it's actually the opposite of what is being claimed. And that is reason we are not as strict with distortion down there

This is the end of the thread really...the OP has not addressed this comment, which is a rather large oversight. Hearing is not only non-linear with respect to perceived frequency response, it is non-linear as the amplitude changes. Further, we have decades of real-world experience showing how insensitive we are to distortion at very low frequencies.

 

[RCAguy]

Thanks for a thoughtful response!
As I said originally, while I found the approach interesting (and definitely applaud the effort to formalize it into a paper), I still have two main reservations:

1) The "sounds like" wording might be misleading to casual readers who might not catch that the hypothesis is not backed by formal listening tests

2) The equal-loudness contours (whether the older FM or the newer ISO curves) might not be the best target for normalization in this case since they do not take into account auditory masking, which neccessarily comes into play when two or more tones play simultaneously (like a fundamental and its harmonic). It would IMHO be interesting to try and normalize THD to some of the masking curves reported in earlier studies. But even so, formal listening tests would be needed to see how figures obtained in that way relate to audibility.

As it is I unfortunately don't find the hypothesis of inflated perceived LF distortion convincing enough to invest more time looking into it. But I'd be happy to change my mind should enough evidence come to light!

1 Our terminology is defined in the paper to clarify for a broader readership perceived Loudness (“sounds like” in dB phons) from actual SPLdB. But no one here has bothered to read the paper.

2) Masking is certainly valid, but beyond the scope of a hypothesis paper that challenges conventional wisdom and entices listening tests. Compared to highly “produced” (in many cases highly distorted) electric pop music, masking is far lower with a acoustic music. Those listeners - some who originated the hi-fi hobby of the 1950’s - have remembered references for acoustic sounds heard live at unamplified classical & jazz concerts.

 

[RCAguy]

This is the end of the thread really...the OP has not addressed this comment, which is a rather large oversight. Hearing is not only non-linear with respect to perceived frequency response, it is non-linear as the amplitude changes. Further, we have decades of real-world experience showing how insensitive we are to distortion at very low frequencies.

Addressed in #13. I’ve said I welcome discussion specifically regarding ,and with anyone who has actually read, the paper in the OP.

 

[olieb]

But no one here has bothered to read the paper.

You keep stressing this and of course there is a problem in talking about things one has not really seen. But to get a permit and to go and download and read one needs an incentive, at least that is the case for me. Somehow you do not seem to bring that to the table.

The way I see it you present an unusual and rather interesting way for looking at distortion figures. Instead of using "objective" quantities like power or SPL you use the perceived loudness from ISO226 as the base figure. Thanks for the diagram BTW.

It is interesting because it shows that the "loudness" of a harmonic distortion might be quite a bit higher than the usual figures would suggest.
With a fundamental at 30 Hz at 90dbSPL (≈40 phon) the third harmonic (H3 - 90Hz) with 3% (-30db -> 60dbSPL) would have loudness of a bit over 30 phon.
That looks like a lot of distortion, but it is the very same as before of course, it is just a different way of describing the situation. Nothing has been "inflated". 31 phon HD3 with a 30 Hz fundamental of 40 phon sounds exactly like 3% at 90dB because is is the same thing. And for all we know it is not audible.

So it is not clear - at least to me - what is the "hypothesis"? I do not see it, no tip of the iceberg there. And therefore there is little incentive to work through the paper. The fact that it is advertised for 5$ adds to the reserve, seems a bit off to me.

And then there is the point of masking which obviously is the main issue with the perceptibility of distortion.
The equal loudness curves work in a way that harmonic distortion in low frequency might be perceived louder than "expected". But the experience tells us that nevertheless even high distortion is not a big problem. So masking seems to increase even more strongly towards low frequency than the mentioned loudness tendency works in the other direction.
So without taking masking into account there probably cannot be a meaningful hypothesis.

And in the end we know the result already, don't we? HD in low frequency is audible only to a very small degree.
I do not see a reason to discard that as "conventional wisdom" that has to be overcome. At least not without good experimental evidence.

 

[Curvature]

@RCAguy I paid for the paper. Issuu has to be worst host. I can't download the paper, and I can only read it on its portal, only in landscape mode. Add to that the paywall and required account creation.

With respect, your paper is poorly written. For your main claim--you can't equate the compression of loudness at LF to "perceived inflated THD". The curve compression shows greater sensitivity to SPL changes, but it does not mean that the distortion is audible. Audibility is not the same as sensitivity.

Here's a well known study on LF distortion with masking taken into account: https://secure.aes.org/forum/pubs/journal/?elib=5147

The table shows how high a harmonic has to be before it begins to be perceivable. Take the 50Hz result for the 2nd harmonic at 80dB SPL: -37dB or around 1.4%, or 43dB SPL at 100Hz. That puts it roughly on the 20 phon curve, which means that the distortion will be as loud as 20db SPL at 1kHz. That's all that means, though. You still only have competing tones of 50Hz vs. 100Hz, nothing else. And then with music this will be very moot. Music is an excellent masker of its own low-level content and the problems of loudspeakers.

Here's the masking curves for 50Hz at three SPLs ("hearing acuity" is 0dB phon):

I understand the desire to pass on knowledge but your paper is not how it is done. I agree that less distortion is better in speakers, that not using LPF/HPF for sub vs. mains is a poor idea, particularly in a PA context.

Your paper has many other problems. That's another way of saying you simply did not do enough hard research and jumped to conclusions.

If I were you I would take your paper down. Other people should not read it and be confused by it.

 

[fineMen]

@RCAguy I paid for the paper.

Thank you so much, especially for the time it may have cost you!

I agree fully with your comment in every detail, including the "confusion" part.

I only want to underline that, in ironic contrast to the here raised problem statement, the experimentally found threshold for distortion audibility increases with decreasing volume (Table 2). If the statements by RCAguy were true, the opposite would be (kind of) expected. Reiterated, the musings are based on a fundamental fallacy--twice.

Thank's again!

 

[RCAguy]

@RCAguy I paid for the paper. Issuu has to be worst host. I can't download the paper, and I can only read it on its portal, only in landscape mode. Add to that the paywall and required account creation.

With respect, your paper is poorly written. For your main claim--you can't equate the compression of loudness at LF to "perceived inflated THD". The curve compression shows greater sensitivity to SPL changes, but it does not mean that the distortion is audible. Audibility is not the same as sensitivity.

Here's a well known study on LF distortion with masking taken into account: https://secure.aes.org/forum/pubs/journal/?elib=5147

The table shows how high a harmonic has to be before it begins to be perceivable. Take the 50Hz result for the 2nd harmonic at 80dB SPL: -37dB or around 1.4%, or 43dB SPL at 100Hz. That puts it roughly on the 20 phon curve, which means that the distortion will be as loud as 20db SPL at 1kHz. That's all that means, though. You still only have competing tones of 50Hz vs. 100Hz, nothing else. And then with music this will be very moot. Music is an excellent masker of its own low-level content and the problems of loudspeakers.

Here's the masking curves for 50Hz at three SPLs ("hearing acuity" is 0dB phon):
View attachment 308902

I understand the desire to pass on knowledge but your paper is not how it is done. I agree that less distortion is better in speakers, that not using LPF/HPF for sub vs. mains is a poor idea, particularly in a PA context.

Your paper has many other problems. That's another way of saying you simply did not do enough hard research and jumped to conclusions.

If I were you I would take your paper down. Other people should not read it and be confused by it.

Thanks, Curvature, for the well said critique of the paper, at least part of which you are the first in this group to have actually read. Per your advice, I have taken it down.

As a 40yr member of AES, I am thoroughly familiar with the AES paper you cite, co-written by a personal friend, Eric Benjamin. However it was written (with the vast resources of Dolby, which is in the "percentual masking" business) in 1988 based on the original F-M contours, not the updated ones in ISO226:2003\2023, which are quite different, far steeper, requiring in my view a fresh look. Nevertheless, I conclude in 2023 the same as Fielder & Benjamin in 1988: “Based on all the frequencies and levels surveyed resulted in the requirement that the second harmonic be less than 3%, the third harmonic be less than 1%.” Even with strides made since 1988, few subwoofers achieve these criteria.

Most points you raise have been discussed in this thread, so I needn't respond further. I will emphasize re masking that highly-produced popular music, with its relentlessly loud levels and already copious harmonics, often from intentional clipping, are not the whole picture of audio. I've practiced for decades recording acoustic sounds (classical music, jazz, chorale, film production sound) and mixing hundreds of films & TV programs where levels are often soft, and harmonic and IM content is far less "busy." For these genre, distortion is far less masked. So the premise of the paper is that lowering distortion is a good thing - it can always be added to taste by inserting a custom-biased 12AX7. A recent case in point is attending "Oppenheimer" in an IMAX theater, which are operated significantly louder than SMPTE standard, and where LF effects and music suffer, unmasked, from SW distortion (drivers, post crossover).

I feel we can be glad that manufacturers such as JBL, Monoprice, and Rythmik etc (examples analyzed in the paper) keep introducing lower and lower distortion subwoofer models. Why do they persist? Are they fools for doing so when the artifacts may be\might not be masked?

 

[phoenixdogfan]

So, what exactly is 'perception'? If the distortion component was singled out, in that the base tone (that what undergoes the distortion) is taken away but the distortion component is kept, then your musings are correct, but only tell what the F/M curves are about.

If you insist that the base tone is kept and heard with the distortion component together, the prerequisites / conditions for the F/M curves are no longer met. The F/M curves are for single tones, not for two or even more combined.

What chimes in with the combination of tones is the masking effect (s/ above many times). So regarding distortion you are in an entirely different regime. This mode as to say was also investigated by scientists in the field. We all know the results, which, to our all relief, indicate that the bass doesn't ask for that much of a concern regarding HD.

Hand your paper in to the Audio Engineering Society, really, they are keen on new stuff. Then you'll see if they give you an answer.

We already have psychoacoustic studies which indicate the audability of HD with music is inversely related to frequency because of the masking effect.

How Much Distortion Can We Hear With Music?

How much distortion can we hear with music? An experimental study at Axiom.

www.axiomaudio.com

 

[fineMen]

... hundreds of films & TV programs where levels are often soft, and harmonic and IM content is far less "busy." For these genre, distortion is far less masked.

Alas, another misconception. Two things at work here:

1. masking by other content, and if it was the harmonic series of the instrument alone ( was not addressed here in )
2. masking by the base tone, that tone which provokes the harmonics in the loudspeaker ( was addressed many times here in )

You address the first, while with the second you could make a point. In case the fundamental tone is too low in amplitude to be heard, then the harmonics would stand out, for sure. But what type of a sub would that be, generating harmonics, but the fundamental is not even audible? We're coming closer to that with increasingly smaller speakers. Because the general publih isn't interested in stereo etc as a status icon anymore. It's more the vegan bratwurst and the newest smart phone, the latter with stereo speakers and stereo lenses. Tides have turned.

 

[Curvature]

Thank you so much, especially for the time it may have cost you!

I agree fully with your comment in every detail, including the "confusion" part.

I only want to underline that, in ironic contrast to the here raised problem statement, the experimentally found threshold for distortion audibility increases with decreasing volume (Table 2). If the statements by RCAguy were true, the opposite would be (kind of) expected. Reiterated, the musings are based on a fundamental fallacy--twice.

Thank's again!

I wanted to clarify the data in Fielder's study.

Here are the relative masking thresholds from 10, 20, 50 and 100Hz, at 80, 100 and 110dB SPL, for H2-5 distortion.

Same as above, except absolute thresholds (actual SPL).

There are distinct changes in slope for higher harmonics. I am not sure how to interpret the data further. The curves are not always consistent. As far as I understand, above bass frequencies, the usual tendency is for masking curves to become sharper (higher Q) above 80dB SPL, making distortion more audible rather than less. There are not many sources of masking curves or other information below 200Hz or so. And then there will be some variability of listeners' hearing which will make curves have a different shape.

Attached are the mdats and other related files.

 

[fineMen]

I wanted to clarify the data in Fielder's study.

Here are the relative masking thresholds from 10, 20, 50 and 100Hz, at 80, 100 and 110dB SPL, for H2-5 distortion.
View attachment 309602

The curves that originate in point 50Hz, 0dB. Assumed they depict the threshold for detecting an additional tone at frequency x, it still appears to me, that at higher volumes the masking is more effective violet (110) > cyan (100) > orange(80). That would support the idea of F/M inflating the relevance of distortion. Got it, thank you. The number table above conveyed the opposite.