Do we crave distortion?

[lashto]

You do not seem to understand masking ..

That would be easy to debate. But also quite a waste of time ...

Masking can, at best, explain some of the experimental results.
It may explain why some HD-spectra are more/less audible (or more/less objectionable.)
But it cannot explain why some HD-spectra are both audible and preferred.

OTOH, the ear-HD spectrum can explain both aspects: more/less audible and more/less prefered.
My first reaction when I saw that graph was like "hm, that looks a lot like a tube HD graph ... particularly the 300B". That alone may explain quite a few preferences. Some may even consider it enough for a q.e.d. Some not...

 

[DonH56]

So the argument is that, if the ear adds distortion, then adding more distortion in similar fashion must be pleasing?

An alternative theory, for which I have no proof (nor care to find any), is that our brain compensates the ear's transfer function and adding additional distortion is less pleasing.

 

[lashto]

Distortion implies no longer true to the source material. Your examples are of modifying the source, not necessarily adding distortion from our own senses.

You can see it that way too. Not sure how it changes my arguments though.

And I'd like to see the reference for the study that you linked; most (not all) of the ones I am familiar with have been debunked over the years. In any event it does not violate my premise that distortion in audio is preferred by some listeners, albeit work by Olive and Toole show the opposite. It's usually pretty easy to find studies on both sides of any given issue, though I have no desire to go digging for them. Or to read this thread more than a post or two every other week or so.

AFAIK, Olive & Toole just (re)demonstrated that THD-numbers are not corelated to preferences. And that FR-linearity and controlled dispersion are preferred. Not debating any of that.
But when it comes to HD, they pretty much ignored it. I only found stuff like "more HD studies are needed" and "less important than FR/etc". Olive may have changed his mind lately.

I linked many studies in this thread. Lots here.
Geddes shows a clearly (albeit slightly) negative/inverse relation between HD/IMD and preferences: see Figure 2 and 3. And his diyaudio thread is full of statements like "studies of mine have clearly shown the human PREFERENCE for distortion" (his own caps).
And the latest I linked, comes from University of New Hampshire.

I'd also like to know more about those "have been debunked" studies. Got some links? Or at least titles/authors/etc that I can google?

 

[Axo1989]

Looks like Olive changed his mind lately.

I maybe missed that thread: Olive liking GoldenSound's preso? Cue chapter 5 of the War on Cameron !!

Edit: or not, looks like our host didn't dive in ...

 

[lashto]

So the argument is that, if the ear adds distortion, then adding more distortion in similar fashion must be pleasing?

I guess we can agree that "more of the same" would be less objectionable (and less audible).
More pleasing is a bigger 'stretch'. But why not?

An alternative theory, for which I have no proof (nor care to find any), is that our brain compensates the ear's transfer function and adding additional distortion is less pleasing.

That sounds reasonable too. But I've been asking the whole week for a single study that says "additional distortion is [always] less pleasing". Still waiting...

In the meantime, TheHypothesis: "more of the same" (ear-like-HD) can sound pleasing. Up to an unknown limit/level.

Predictions based on the hypothesis:

1. any device which adds "more of the same" will sound pleasing (or at least not bad).
2. the closest to ear-HD, the more pleasing.
3. when comparing 2+ devices, the most ear-like-HD-spectrum wins the preference-contest (as always, everything else should be ~equal)
4. more ear-like-HD will win the preference-contest against less (up to a limit)

Tube devices add "more of the same" HD. Millions of people like/buy them. Check (P1)
Turntables add "more of the same" HD. Millions of people like/buy them. Check (P1)
Studios add "more of the same" HD to music. Billions of people like/buy it. Check (P1)
The closest (I know) to ear-HD-spectrum is the 300B tube. It's also the most popular. Check. (P2)
New Hampshire study. Check (P3)
@Archimago @pkane added ear-like-HD to some music tracks and organized a large DBT. People prefered the extra "more of the same" (and were able to hear it!). Check (P3/4)
Other studies I linked found extra cases of "more of the same" = "more pleasing". Check++

Spoiler: more..

... may use this space to add more examples later ...
Some businesses (e.g. Nelson Pass') seem to be built upon the presumption that "more of the same" sounds good. Many customers agree. Check (P1)

Spoiler: more far-fetched stuff

F-noise seems to be a kind of "more of the same" (very similar looking downward spectrum). Sounds pleasing, suposedly it's relaxing and the most pleasing 'noise'. It's also part of many natural phenomena: e.g. heart beats and neural activity have that spectrum. Music has that spectrum too .. even non-human music (whatever that is).
What if that "more of the same" enhaces neural activity?! I surely feel more 'stimulated' by a tube amp.
Or maybe it stimulates whatever harmonic-pleasure-center..

Anyway ... IMO the hypothesis can (already) be called a theory!

Counter-examples highly welcome!

 

[lashto]

I maybe missed that thread: Olive liking GoldenSound's preso? Cue chapter 5 of the War on Cameron !!

Edit: or not, looks like our host didn't dive in ...

he also doesn't seem to like ASR much. The horror

I do like ASR. I think it's the best and most useful audio forum. By quite a margin!
But I also think that many here are quite wrong about 'distortion' ... and anyone who wants to prove the opposite is highly welcome.

 

[Axo1989]

Tube devices add "more of the same" HD. Millions of people like/buy them. Check.
Turntables add "more of the same" HD. Millions of people like/buy them. Check.
Studios add "more of the same" HD to music. Billions of people like/buy it. Check.
The closest (I know) to ear-HD-spectrum is the 300B tube. It's also the most popular. Check.
The studies I linked found quite a bit of "more pleasing". Check.

We'd have to establish whether people are liking more of the same, or simply not hearing it. A little while back, @GXAlan posted three sets of comparison clips from a 300B SET amp with high THD+N measurements (edit: measuring 22 dB SINAD, but equivalent to 35-40 dB SINAD at the level recorded, see post below) vs a very low distortion Topping model. Almost no-one participating here could reliably differentiate blind. A few people could, and I could hear the harmonic distortion as a kind of ringing sound (not to be confuse with pre-ringing) most noticeably one one of the tracks but it was quite subtle. The actual tube amp distortion spectra was the usual H2-dominated with the higher orders falling off. On one of the two headphones I tried, the SET amp jazz track was both different and pleasing.

Spoiler: some more far-fetched stuff

F-noise seems to be a kind of "more of the same" (very similar looking downward spectrum). Sounds pleasing, suposedly it's relaxing and the most pleasing 'noise'. It's also part of many natural phenomena: e.g. heart beats and neural activity have that spectrum. Music has that spectrum too .. even non-human music (whatever that is).
What if that "more of the same" enhaces neural activity?! I surely feel more 'stimulated' by a tube amp.
Or maybe it (directly) stimulates whatever pleasure-center..

Well, that relaxing swooshy stuff for sure, and I've posted some other (less relaxing) examples of music with considerable deliberate distortion upthread.

I've not felt the need to add distortion via the reproduction chain, but I've not experimented with doing so either. For some music it may make a difference. But when I had a damaged loudspeaker a while back I had to stop listening to say Lana Del Rey because it just didn't work, but could happily listen to a wide range of alt/grunge/screamo or classics like My Bloody Valentine, because lashings of deliberate distortion masked the poor loudspeaker's misdeeds [see post #376].

Anyway ... IMO the hypothesis can (already) be called a theory!

Counter-examples highly welcome!

It's a theory in the vernacular, but not if we adhere to strict scientific terminology (as another pedant noted upthread). It always helps to be clear on those things.

 

[GXAlan]

A little while back, @GXAlan posted three sets of comparison clips from a 300B SET amp with high THD+N measurements (corresponding to 22 dB SINAD)

Take the blind challenge! 300B SET vs. Straight Wire with Gain

You can get very powerful bass with tube amplifiers if you pay attention to the load impedance where bass energy is made. It is nice to get away from that particular dependency though! WRT your last comment: Yikes! If the SET was really that bad (rolled off in the audio band with a resistive...

www.audiosciencereview.com

And

Record breaker: Raphaelite CS30-MKII tube amplifier review and measurements

Raphaelite CS30-MKII 300B integrated tube amplifier review: A New Performance Record This is a review and detailed measurements of Raphaelite CS30-MKK 300B single ended tube amplifier. It is available on eBay for $1449 + $310 shipping to California + tax. There are some better deals from...

audiosciencereview.com

To be fair, at the recorded level, it was probably closer to 35-40 dB SINAD since it wasn’t at “5W” equivalent

 

[Galliardist]

Can I throw this article into the mix?
Audibility of Small Distortions | madronadigital

It's more about audibility than preference but does have some relevance.

 

[Karmacoma]

When I'm listenning to a guy playing a Stradivarius, no, I'm not gonna add distorsion.
Of course not.
Not open for discussion on the subject.

 

[solderdude]

Harmonics generated by an amplifier are considered as distorsion in world of audio since something is added that was not on the source. But in the world of music, harmonics added to the original makes it sound richer, especially the 2nd harmonics: it is like adding other performers that plays the same instruments and the same notes but one octave higher: our ears like it !

Of course, in case of instruments the harmonics (especially 2nd and 3rd) more often than not are even higher than the fundamental or merely just below that in level... and not -70dB or so.
Below a violin note being played. Anyone who can guess what the addition of any harmonic at -60dB will change to the sound when 2nd harmonics by themselves are already -1dB so 1000x smaller in amplitude ?

This basically means that the 2nd harmonic of this violin (at -2dB) will be played back at -1.9913dB so when you can actually hear a level difference of 0.0087dB you can also hear -60dB in harmonics being added. Good luck with that.
Even the (not so nice sounding ?) 11th harmonic is just down -7dB or so.

Harmonics of instruments have their own (and amplitude/time dependent) spectrum while electronics have a fixed (but amplitude dependent) spectrum and have IM distortion which can be far more audible as it may not be masked and is not harmonic in any way.

One should not confuse harmonics of instruments with harmonics generated by electronics.

 

[Galliardist]

Of course, in case of instruments the harmonics (especially 2nd and 3rd) more often than not are even higher than the fundamental or merely just below that in level... and not -70dB or so.
Below a violin note being played. Anyone who can guess what the addition of any harmonic at -60dB will change to the sound when 2nd harmonics by themselves are already -1dB so 1000x smaller in amplitude ?

This basically means that the 2nd harmonic of this violin (at -2dB) will be played back at -1.9913dB so when you can actually hear a level difference of 0.0087dB you can also hear -60dB in harmonics being added. Good luck with that.
Even the (not so nice sounding ?) 11th harmonic is just down -7dB or so.

Harmonics of instruments have their own (and amplitude/time dependent) spectrum while electronics have a fixed (but amplitude dependent) spectrum and have IM distortion which can be far more audible as it may not be masked and is not harmonic in any way.

One should not confuse harmonics of instruments with harmonics generated by electronics.

The point is though that the harmonics generated by electronics are added to every frequency we hear. In the case of a violin, there are as you show the higher harmonics, and also resonances that have little to do with the fundamental, and they are also affected by the harmonics added by the electronics. That may well result in the apparent sweetening of the sound sometimes reported with addition of second harmonic. It's not distorting the fundamental audibly, but it is distorting the instrument's sound.

I guess at this point it's time to pull out my favourite anecdote from the old HiFi Answers magazine. It refers to blind testing by a turntable designer, who had his test system behind a curtain. A friend called in regularly to hear his latest design, and each week complained that the sound of the violin on the new turntable was wrong. After about six weeks of this, the friend called round again and again heard the violin music. "That's the worst yet" he complained, "nothing like a violin at all".

Out from behind the curtain, stepped the violinist.

This is a problem with blind preference testing, and is the reason why the Harman tests aren't just done with any old recording. I'm not convinced that getting a violinist in is a fair test against a playback system in a small room, because the violin will itself sound different ("My art is meant to be heard at a distance" - Paganini). It does I think throw light on this discussion though!

 

[antcollinet]

and they are also affected by the harmonics added by the electronics

A harmonic of a harmonic is also a harmonic.

That is to say - electronics adding 2nd harmonic to the 2nd harmonic of an instrument - will be adding (say) -60dB to the 4th harmonic of the instrument. With an equivalent inaudibility to that of the addition to the actual instrument second harmonic described by @solderdude.

If that makes sense

 

[Sokel]

Of course, in case of instruments the harmonics (especially 2nd and 3rd) more often than not are even higher than the fundamental or merely just below that in level... and not -70dB or so.
Below a violin note being played. Anyone who can guess what the addition of any harmonic at -60dB will change to the sound when 2nd harmonics by themselves are already -1dB so 1000x smaller in amplitude ?

This basically means that the 2nd harmonic of this violin (at -2dB) will be played back at -1.9913dB so when you can actually hear a level difference of 0.0087dB you can also hear -60dB in harmonics being added. Good luck with that.
Even the (not so nice sounding ?) 11th harmonic is just down -7dB or so.

Harmonics of instruments have their own (and amplitude/time dependent) spectrum while electronics have a fixed (but amplitude dependent) spectrum and have IM distortion which can be far more audible as it may not be masked and is not harmonic in any way.

One should not confuse harmonics of instruments with harmonics generated by electronics.

Yep.
I tried to explain the same,see the trumpet's harmonics in my example,way above the fundamental in my post:

Do we crave distortion?

You dont understand how flawed the stereo system is - there is no perfection in 2 channel playback even If the gear has a SINAD of 120. Its only a bad illusion to be had. I'm really quite happy with the things my lowly stereo system does. It has a solid phantom center, the sound often goes well...

www.audiosciencereview.com

 

[Sokel]

A harmonic of a harmonic is also a harmonic.

That is to say - electronics adding 2nd harmonic to the 2nd harmonic of an instrument - will be adding (say) -60dB to the 4th harmonic of the instrument. With an equivalent inaudibility to that of the addition to the actual instrument second harmonic described by @solderdude.

If that makes sense

The thing is that the electronics will see each harmonic as a fundamental,they not only they will add H2 to H2 but also H2 to H3 and so on.
I don't thing we want that.

 

[antcollinet]

The thing is that the electronics will see each harmonic as a fundamental,they not only they will add H2 to H2 but also H2 to H3 and so on.
I don't thing we want that.

No we don't

But as I point out each of the harmonics of a harmonic are also harmonics of the fundamental. @solderdude 's argument was (I think) that with normal levels of harmonic distortion from an amp the distortion products are tiny compared with the fundamental, and already existing harmonics from the instrument, representing a "difference of 0.0087dB"

So the argument that they can be somehow beneficial to the sound experience is nonsense.

Then all you are left with is IMD which is definitely not.

It is similarly difficult to believe HD could result in "sweetening of the sound". Harmonics of inharmonic resonances will just add to the low level "grass" of IMD

 

[UliBru]

IMO most people believe that a 2n HD will add just double frequencies to existing harmonics. But this is not the case. A 2nd HD also adds IM.
Here is an example. A multitone signal created with 10 frequencies from 23.4375 Hz to 12 kHz (48 kHz samplerate). So the distance between each frequency is exactly one octave = factor 2. The chart looks like this

if the 2nd HD would just double the existing frequencies the created new frequencies would also just cover the existing lines. There would be the only one new frequency line at 24 kHz. Now let's add a 2nd HD of -20 dB by the nice tool DISTORT. The result is

So there are many new frequencies which can be considered as intermodulation distortions. All these new frequencies are not sounding harmonious.
Of course someone can argue that -20 dB is pretty much and discuss about the enjoyment at levels along a just noticable difference (JND).
But I clearly prefer not to add quite arbitrary and unpredictable new frequencies to the original track.

PS: the noise level of the green signal is caused by the 32bit float resolution of the DISTORT result.

 

[antcollinet]

IMO most people believe that a 2n HD will add just double frequencies to existing harmonics. But this is not the case. A 2nd HD also adds IM.
Here is an example. A multitone signal created with 10 frequencies from 23.4375 Hz to 12 kHz (48 kHz samplerate). So the distance between each frequency is exactly one octave = factor 2. The chart looks like this
View attachment 335692

if the 2nd HD would just double the existing frequencies the created new frequencies would also just cover the existing lines. There would be the only one new frequency line at 24 kHz. Now let's add a 2nd HD of -20 dB by the nice tool DISTORT. The result is
View attachment 335693

So there are many new frequencies which can be considered as intermodulation distortions. All these new frequencies are not sounding harmonious.
Of course someone can argue that -20 dB is pretty much and discuss about the enjoyment at levels along a just noticable difference (JND).
But I clearly prefer not to add quite arbitrary and unpredictable new frequencies to the original track.

PS: the noise level of the green signal is caused by the 32bit float resolution of the DISTORT result.

As posted about 5 pages ago by @pkane (post 480)

 

[UliBru]

As posted about 5 pages ago by @pkane (post 480)

Thanks, I have missed to read this post. In Germany we say something like: two fools one thought

 

[antcollinet]

two fools

Not in this case though