So we now have a speaker with >5% distortion which isn't audible subjectively. Is a SINAD of 96dB poor?

[Frank Dernie]

I do also agree with the earlier post regarding high quality engineering. Even if it’s inaudible, it’s interesting to see the art of the possible and I’d much rather give my money to someone who is trying very hard to build an excellent device than someone building just good enough.

I am a mechanical engineer.
It is generally considered foolish to make a product perform better than it needs to, particularly since it almost always is more expensive to do so.
Also adding .001% distortion to 5% is not going to make any difference.

Frequency matters a lot, that >5% is below 500Hz.

While this is not a peer-reviewed study, this is what Axiom Audio found:

5% (-26dB) becomes inaudible when below ~650Hz. That's just detecting it, so when it becomes subjectively harmful would be even higher.

As for DAC, Amir has shown that you can get super high SINAD for relatively very cheap, so there is no need to buy an expensive DAC with low SINAD, unless for some reason who like the looks or whathaveyou.

I am shocked.
Admittedly I mainly listen to classical music but the vast bulk of the music I listen to has a LOT of information below 500Hz, usually more than half of it. It is quite normal for the level at frequencies >2kHz to be 40dB lower than the level at 250 Hz.

 

[MZKM]

I am a mechanical engineer.
It is generally considered foolish to make a product perform better than it needs to, particularly since it almost always is more expensive to do so.
Also adding .001% distortion to 5% is not going to make any difference.


I am shocked.
Admittedly I mainly listen to classical music but the vast bulk of the music I listen to has a LOT of information below 500Hz, usually more than half of it. It is quite normal for the level at frequencies >2kHz to be 40dB lower than the level at 250 Hz.

Well, then you should be happy that humans are less sensitive to distortion as frequency lowers.

 

[RayDunzl]

Or we can make it even quicker if somebody knows of some distortion-injecting filter in Audacity, then we can play until it becomes detectable. All while on a speaker with a distortion higher than that.

Try:

https://www.audiosciencereview.com/...test-distort-audibility-of-distortions.10163/

 

[SIY]

Ok. I wasn't suggestion that amps have predominantly higher-order harmonic distortion, but some certainly do (read your own final comment in this same post )

"Copious" is not "predominant." The 2nd and 3rd are still the highest, but there's large, "copious" amounts of higher order. Much, much more than engineered amps.

 

[wwenze]

This is not perfect for true controlled ABX testing, but not bad, and very simple and convenient.

Very convenient indeed. I gotta read up what "THD IMD" does tho.

The A minus B feature is exceptionally beautiful, since previously I only knew of Diffmaker which had to use two different input files then do some correlation analysis to time-match.

At my max normal listening levels, I can hear things at around 0.5% if I put my ear 15cm in front of the speaker. 0.1% is for practical purposes silent at these listening levels. Makes sense if the distortion components are below 0dB SPL real life.

 

[Frank Dernie]

Although to be fair, the Fostex 'speaker is just the latest incarnation of their 'singing brick' loudspeakers which were everywhere in studios and radio stations for years. They were never intended as anything other than 'signal present' monitors, not quality monitoring, and anyway, measuring distortion at 90dBSPL on something that small is clearly well beyond its design capabilities.

Proper loudspeakers, i.e. those on which one can judge quality, generally have much lower distortion, comfortably under 1%, and at mid and high frequencies, typically 0.1%, maybe 0.2% at 90dBSPL.

S

My point was that Amir didn't find the distortion audible when he listened loud - yes, not a peer reviewed study )) but he does listen loud to the stuff he tests and has commented about the poor performance of the Genelec at high volumes and some others with the woofer bottoming out.
Also, A quick look through the HiFi News measurements at 90dB does show 1kHz and 10kHz (totally uninteresting IMHO) distortion levels of 0.1 and 0.2% it is uncommon and usually on big, expensive and efficient speakers.
IMO the distortion value at 100Hz is more interesting in terms of the audible harmonics it is measuring than the 10hHz one.

 

[andreasmaaan]

"Copious" is not "predominant." The 2nd and 3rd are still the highest, but there's large, "copious" amounts of higher order. Much, much more than engineered amps.

https://www.audiosciencereview.com/...ly-is-a-sinad-of-96db-poor.12519/#post-370859

 

[Frank Dernie]

5% (-26dB) becomes inaudible when below ~650Hz

This just generates another question.
If this really is true we don't, and never have, need a low distortion high fidelity system for listening to music.
Maybe it is true that only frequency response matters.

 

[RayDunzl]

If this really is true we don't, and never have, need a low distortion high fidelity system for listening to music.

I always swap to my low distortion high fidelity speakers when listening to music.

For general TV and not paying attention listening, the little distortion boxes are fine.

 

[MZKM]

This just generates another question.
If this really is true we don't, and never have, need a low distortion high fidelity system for listening to music.
Maybe it is true that only frequency response matters.

It’s for sure the the vast majority of what matters.

 

[wwenze]

This just generates another question.
If this really is true we don't, and never have, need a low distortion high fidelity system for listening to music.
Maybe it is true that only frequency response matters.

Left - Piano, Right - Oboe, Bottom - Voice, Source - Here

A small % of THD from reproducing lower frequency components is not going to change the shape by much. A change in frequency response will.

 

[Frank Dernie]

Left - Piano, Right - Oboe, Bottom - Voice, Source - Here

A small % of THD from reproducing lower frequency components is not going to change the shape by much. A change in frequency response will.

Indeed, but the timbre of a particular instrument is defined by the ratio of the harmonics to the fundamental so a distortion of 5% may well change the timbre of an instrument enough for the player to notice it even if a record purchaser wouldn't since they are unaware of the original timbre and just still recognise a piano, or whatever, but not notice it hasn't got the usual Steinway timbre, for example.

Non-flabby bass and accurate instrumental timbre have always been the determining factor on whether I chose a speaker or not and I have found some well regarded speakers to unacceptably colour the timbre for me, but then common frequency response waviness is probably big enough to mask distortion in altering the magnitude of overtones and that is all there is to it.

 

[sergeauckland]

This just generates another question.
If this really is true we don't, and never have, need a low distortion high fidelity system for listening to music.
Maybe it is true that only frequency response matters.

There are plenty of people who deliberately choose high distorting systems like vinyl, SET valves and horns, claiming they sound more real, better than low distortion equipment. I don't suppose the frequency response is flat either!

S.

 

[ROOSKIE]

I am a mechanical engineer.
It is generally considered foolish to make a product perform better than it needs to, particularly since it almost always is more expensive to do so.
Also adding .001% distortion to 5% is not going to make any difference.


I am shocked.
Admittedly I mainly listen to classical music but the vast bulk of the music I listen to has a LOT of information below 500Hz, usually more than half of it. It is quite normal for the level at frequencies >2kHz to be 40dB lower than the level at 250 Hz.

I have discussed this in several speaker review threads. This is not new. There are have been discussions around speaker distortion for as long as I have been interested.
Most people are very surprised at first to discover that when speakers themselves distort it is hard to hear. (electronics are another story) The general consensus around what speaker distortion levels cause obvious problem does not exist.
This is especially true as THD is made up of all kinds of distortion. 5% of one type of distortion is not the same as 5% of another. Generally I found a fairly stable consensus around 2nd and 3rd order harmonic distortion in speakers. The consensus is that it is not really a problem even at whole percentages. Some of it is even desired as it may be a big factor in what makes the speakers sound warm (2nd order) or have a pleasing/not harsh edge or slight crispness (3rd)
A lot of this discussion has taken place in DIY forums and on AVS and other audio sites. I can't track links down but the info is out there. In some cases a person would not notice even 100% THD such as in very low frequencies.
I remember Tom Nousane did a number of studies that seem to no longer be on the web. From my memory most listeners (in bind tests) did not notice 25% distortion in bass and upper bass frequencies (I remember one person could hear it at 10%). My understanding is that it gets easier as you climb up into the mid-range and highs.
Here is one link to an interesting study speaker manufacturer Axiom did. just 8 or 10 people though.
https://www.axiomaudio.com/blog/distortion/

 

[briskly]

Maybe this can be a starting point. Let's see how much I have to clip before I start hearing stuff.

Nonlinear modeling in the digital domain can produce artifacts that would exceed Fs/2. The artifacts will wrap around Fs/2 back into the intended band, aliasing. Aliasing errors will not fit simple HD testing.

what "THD IMD" does

It could be a Taylor series model of nonlinearity. Harmonic distortions and intermodulations fall out very simply. This model is blind to the frequency dependence of nonlinearity.

Admittedly I mainly listen to classical music but the vast bulk of the music I listen to has a LOT of information below 500Hz, usually more than half of it. It is quite normal for the level at frequencies >2kHz to be 40dB lower than the level at 250 Hz.

At higher frequencies, the decreasing amplitude will be offset in part by the increased number of harmonics. For music, there are usually enough harmonics that the power spectrum coarsely follows 1/f (pink noise, equal energy per octave) over a broad bandwidth.
Another note: the music tracks used to detect nonlinear distortion tends to run inverse to detecting linear distortions. The harmonically rich music tracks that tend to reveal linear errors well are not the minimally harmonic content needed to identify nonlinear errors.

 

[ROOSKIE]

There are plenty of people who deliberately choose high distorting systems like vinyl, SET valves and horns, claiming they sound more real, better than low distortion equipment. I don't suppose the frequency response is flat either!

S.

Real life is very distorted. If you like live shows that are amplified what do think the chances are that the systems they use have low distortion and flat response?
While I determined I tend to like "accurate sound" and "tight sound" and "clean sound" and "studio style sound" more often, this is not always the case and deff not because I think it is more real. I just like it.

 

[Frank Dernie]

(electronics are another story)

Why?

 

[wwenze]

Indeed, but the timbre of a particular instrument is defined by the ratio of the harmonics to the fundamental so a distortion of 5% may well change the timbre of an instrument enough for the player to notice it even if a record purchaser wouldn't since they are unaware of the original timbre and just still recognise a piano, or whatever, but not notice it hasn't got the usual Steinway timbre, for example.

Non-flabby bass and accurate instrumental timbre have always been the determining factor on whether I chose a speaker or not and I have found some well regarded speakers to unacceptably colour the timbre for me, but then common frequency response waviness is probably big enough to mask distortion in altering the magnitude of overtones and that is all there is to it.

So it becomes a mathematical problem. Say you have the second harmonic at roughly -10dB of the fundamental. Add 5% or -26dB from the HD of the fundamental, it is still -10dB. Change 1dB in terms of frequency response on the other hand...

 

[Frank Dernie]

Real life is very distorted. If you like live shows that are amplified what do think the chances are that the systems they use have low distortion and flat response?
While I determined I tend to like "accurate sound" and "tight sound" and "clean sound" and "studio style sound" more often, this is not always the case and deff not because I think it is more real. I just like it.

Well the live shows I go to tend to be orchestras and not amplified but with ear bleeding level rock concerts the sound includes the distortion, why should reproduction add more?

 

[Rick Sykora]

Yes, there does appear to be a major contradiction here. Like many of you, I grew up hearing about how vanishly low distortion (<.1 % THD) made for better electronics. Then to recently see 2% distortion in the sensitive midrange of a highly regarded speaker (that is supposed to be disregarded as inaudible) just seems wrong.

Yes, I get masking, THD measurement shortcomings, psycho-acoustic studies, etc., but whether the amp (or anything else is adding distortion in the chain), why should I accept 2% (or more) from a speaker and accept that my electronics need to be orders of magnitude better?

Although my engineering background is not the same as @Frank Dernie, agree that better quality generally creates more product cost regardless. Clearly getting loudspeakers to electronics distortion numbers is more expensive, but an apparent double standard is not helpful either.