[Poll] Is no-distortion really better than any distortion?

[bobbooo]

P.S.
I find it quite encouraging that we moved from "the numbers do not show D crickets" to "the test methodology was wrong". Real progress

Have you considered that both are actually the case? Discrimination/preference studies are pretty much useless unless they are statistically robust and rule out random chance. I suggest you read this to learn why.

 

[MRC01]

AFAIK there is no 'aparatus' in this world that can fully reverse HD. Theoretically (math) you can apply an 180 degree reversed H2 to cancel the existing H2. But that's only theory.

What you describe seems like more than theory, that sounds like negative feedback.

 

[lashto]

Dear @ lashto we all enjoy some added distortion but that was not the question. The question was if we like it in the reproduction chain.

dear @ tomtoo, that is exactly what the 'trouble' DBT tested. HD was added _after_ the music was done and finished "as the artist wanted it". What archimago added into that file is (approx) the HD spectra of an excellent tubeamp.
It does not matter much if that HD was added by your "reproduction chain" or by Archimago's PC or by lashto's bicycle. It's the same H2, H3, etc. And same as beautiful, wanted, preferred, hated...

 

[lashto]

What you describe seems like more than theory, that sounds like negative feedback.

It does, yes. But negative feedback does not work 100%. Also adds new HDs and kinda plays whack-a-mole on the HD spikes . I'd describe negative feedback as "forced correction" and not a true "chirurgical removal" of HD.

Who knows, maybe someone will invent a device that cleanly removes enough of that HD (as in the remaining is below audible). 100% removal is not possible AFAIK, it's non-linear after all.

 

[tomtoo]

and that is exactly what that DBT tested. HD was added _after_ the music was done and finished "as the artist wanted it". What archimago added into that file is (approx) the HD spectra of an excellent tubeamp.
It does not matter much if that HD was added by your "reproduction chain" or by Archimago's PC or by lashto's bicycle. It's the same H2, H3, etc. And same a beautiful, wanted, preferred, hated...

Ok could you give me please a link to the whole studie??

 

[lashto]

Ok could you give me please a link to the whole studie??

oh well, that genie just did not want to stay in the bottle.

the test. The zip file is still there, highly recommended that you first ABX it yourself.
https://archimago.blogspot.com/2020/01/internet-blind-test-is-high-harmonic.html

the results (that's where I quoted the HD-prefered-graph from). This is just part-II, there are actually III separate analysis articles, all very good reads.
https://archimago.blogspot.com/2020/06/blind-test-results-part-ii-is-high.html

I'll have a beer after all this 'work'. Cheers everyone!

 

[tomtoo]

oh well, that genie just did not want to stay in the bottle.

the test. The zip file is still there, highly recommended that you first ABX it yourself.
https://archimago.blogspot.com/2020/01/internet-blind-test-is-high-harmonic.html

the results (that's where I quoted the HD-prefered-graph from). There are actually III separate analysis articles, all very good reads.
https://archimago.blogspot.com/2020/06/blind-test-results-part-ii-is-high.html

I'll have a beer after all this 'work'. Cheers everyone!

Cheers bro!

I have to read all this first. But cool if you can replace tubes with some math!

 

[solderdude]

I don't quite understand the part about non-reversibility of distortion. Adding noise seems non-reversible. But distortion seems reversible.

Music is never 1 frequency and frequencies are never multiples. You can see what happens in multitones. These are not related frequencies. The IM and HD products are what makes up the 'grass' in the bottom of the plot. And these are with fixed frequencies so the analyzer has time to filter out noise. How would you go about removing frequencies that are all over the place ?
You cannot simply add 'anti distortion' you would have to seriously analyze the transfer function and pull out a lot of math.
It's not like a tone control works at all.

If it were that easy then there would be programs that analyze the music and remove all nasties including those put in there by the creationist making the music.

With linear distortion (say someone pushed up the treble upwards) then this doesn't generate any harmonics. You just push down the treble with a very similar filter and all is original.

Of course adding noise is also 'non repairable' in a perfect sense but there are smart ways to lower noise again but that's another story and has some other but perhaps less objectionable side effects.
Remember Phillips DNL for instance.

 

[tomtoo]

Music is never 1 frequency and frequencies are never multiples. You can see what happens in multitones. These are not related frequencies. The IM and HD products are what makes up the 'grass' in the bottom of the plot. And these are with fixed frequencies so the analyzer has time to filter out noise. How would you go about removing frequencies that are all over the place ?
You cannot simply add 'anti distortion' you would have to seriously analyze the transfer function and pull out a lot of math.
It's not like a tone control works at all.

If it were that easy then there would be programs that analyze the music and remove all nasties including those put in there by the creationist making the music.

With linear distortion (say someone pushed up the treble upwards) then this doesn't generate any harmonics. You just push down the treble with a very similar filter and all is original.

Of course adding noise is also 'non repairable' in a perfect sense but there are smart ways to lower noise again but that's another story and has some other but perhaps less objectionable side effects.
Remember Phillips DNL for instance.

Maybe just maybe with enough math power you could analyze a file and subtract thd? Yes also a ki is just math power.
I think.its always done to reconstruct distorted signals?

 

[solderdude]

But then you would have to know which frequencies and amplitudes are by-products from which transfer function and this would not cure effects not cause by the transfer function.

Nah, the difference is FR is easily corrected, adding noise and frequencies that are not supposed to be there is not easy.
And why would one want to use a tube amp, then apply the opposite before the amp to get not-distorted sound and remove tube goodness with a lot of computation power and trickery when you can just as easily use a good amp that doesn't distort.

Besides with the audiophile tube amps the coloration is never solely determined by a certain transfer function but also load dependent combined with a rarely purely resistive output impedance as well as saturation effects and power supply effects. Try to emulate that or undo it to make a tube amp sound like a SS.

 

[tomtoo]

But then you would have to know which frequencies and amplitudes are by-products from which transfer function and this would not cure effects not cause by the transfer function.

Nah, the difference is FR is easily corrected, adding noise and frequencies that are not supposed to be there is not easy.
And why would one want to use a tube amp, then apply the opposite before the amp to get not-distorted sound and remove tube goodness with a lot of computation power and trickery when you can just as easily use a good amp that doesn't distort.

Besides with the audiophile tube amps the coloration is never solely determined by a certain transfer function but also load dependent combined with a rarely purely resistive output impedance as well as saturation effects and power supply effects. Try to emulate that or undo it to make a tube amp sound like a SS.

No no i not like to talk about tube amps. Imo they are just ir lamps with sound. But it would be interesting to look at older recordings in a different way?

 

[RayDunzl]

AFAIK there is no 'aparatus' in this world that can fully reverse HD. Theoretically (math) you can apply an 180 degree reversed H2 to cancel the existing H2. But that's only theory.

Whenever there is something that "can't be done", I like to see if I can do it, if it interests me, and I have the tools.

so, I did it.

You have to get the phase right, which will vary depending.

---

Using REW, send a 450Hz tone at -20dB to the system (JBL LSR 308 speakers), take an RTA with UMIK-1.

Note high level of 3rd harmonic distortion in the speaker output.

It is clearly audible.

Using the Distortion controls, add third harmonic to the signal at -40 degree phase angle.

https://www.audiosciencereview.com/forum/index.php?threads/distortion-cancellation-experiment.1843/

 

[RayDunzl]

Having just now devoted at least the efforts of an entire neuron to the problem, I come to this hand-waving conclusion.

Harmonic distortion:

If you can measure it, you can remove it.

Easier said than done (for me) but, there it is.

 

[Blumlein 88]

Having just now devoted at least the efforts of an entire neuron to the problem, I come to this hand-waving conclusion.

Harmonic distortion:

If you can measure it, you can, remove it.

Easier said than done (for me) but, there it is.

I don't think I ever posted the results. I've also added reverse distortion to undo it. Possible thanks REW showing phase of distortion. Also done for the speaker. What i found however was if I moved the microphone I had to make new adjustments to counter distortion. So it was somewhat position dependent.

Now if Ray can remove ceiling fan induced doppler distortion we'll be getting somewhere.

 

[RayDunzl]

I don't think I ever posted the results. I've also added reverse distortion to undo it.

I don't claim the invention.

 

[RayDunzl]

What i found however was if I moved the microphone I had to make new adjustments to counter distortion. So it was somewhat position dependent.

It originates at the transducer (as far as audibility in the air is concerned) so, in this case, that is where the measurement would have to be made.

And quickly.

I suppose a proper feedback loop would do.

Easier said than done.

Rythmik uses an anti-coil at the voice coil.

That wouldn't necessarily detect harmonics from a floppy driver flapping.

 

[Blumlein 88]

It originates at the transducer (as far as audibility in the air is concerned) so, in this case, that is where the measurement would have to be made.

And quickly.

I suppose a proper feedback loop would do.

Easier said than done.

Rythmik uses an anti-coil at the voice coil.

That wouldn't necessarily detect harmonics from a floppy driver flapping.

I was measuring at 1 meter. And lateral movement would change it. So I think it was a floppy driver where two parts of it were going different directions. It is probable further away movement would be less of an issue.

 

[lashto]

I don't claim the invention.

Hey, you did steal my HD-cancelling invention
Really cool that you actually tried it, thank you!
The math is good but I thought that in practice there will be some visible residual left. Your result looks waaay cleaner than I expected

 

[RayDunzl]

I was measuring at 1 meter. And lateral movement would change it. So I think it was a floppy driver where two parts of it were going different directions. It is probable further away movement would be less of an issue.

My first guess would be the room intruding on micro-measurements of the phase relationship of tones...

Consider the cancellations in an unsmoothed response at the listening position. Any change in the mic position will change the phase relationship of the direct and reflected and change the cancellations.

The driver isn't making that mess, the room reflections are creating the interference pattern. Move the mic, different set of cancellations.


1/6th and unsmoothed response - same capture data, at 10 feet.

 

[Blumlein 88]

Hey, you did steal my HD-cancelling invention
Really cool that you actually tried it, thank you!
The math is good but I thought that in practice there will be some visible residual left. Your result looks waaay cleaner than I expected

Ray did this some time ago. So it's prior art. As is the ceiling fan jitter discovery.