Do we crave distortion?

[lashto]

You are still mixing production with reproduction.
Production: effects - artistic process. If done well it sells.
Reproduction: effects - bad.

sorry, no mixing here.
Just haven't seen the slightest trace of proof for the "reproduction - bad" part. AFAICS, that is some sort of "objective legend" ... repeated ad-nauseam too

Besides if H2 is sooo great why don’t producers are adding even more of it and save us the trouble adding it randomly during reproduction. It would sell even better according to you.

because "too much of a good thing" is usually not a good thing...
Btw, if anyone wants to get rid of distortion: AI may be able to do that for you (soon). https://crowdunmix.org/the-best-way-to-recover-audio-distortion-using-ai/

.. there are composers who have deliberately broken with the science of harmony (I hope I got the English expression right). And even what might be taboo and odd sounding today, might be en vogue tomorrow.
That's art and rightfully so.
But not engineering.

True and a very good point!
And here we are talking about preferences: whether people like D or not, whether they think that D sounds 'good' or 'bad'.
That is also not engineering!

 

[solderdude]

because "too much of a good thing" is usually not a good thing...

Can I ask you what you think is too much and what would be the minimum required amount of distortion (in dB or percentage) ?

 

[lashto]

here's how a good engineer looks when he discovers that distortion doesn't actually sound bad
(not my image, just a crosslink from Audioholics - Stop Chasing SINAD Distortion in Audio Amplifiers)

 

[lashto]

Can I ask you what you think is too much and what would be the minimum required amount of distortion (in dB or percentage) ?

of course you can. Doesn't mean that I can provide a good/useful answer

It seems to depend on a million things: what kind of distort-spectrum, whether it is dac/amp/speaker distort, the kind of signal/music, listener's preferences/experience/mood etc...

Can only post a personal anecdote: most tube amps do sound good to me (in the right setup: e.g. super efficient speakers, high-impedance HPs etc..). Seen measurements for some of the ones I like, they were in the -40-60dB THD ballpark. Does that mean anything? Probably not much .. especially since THD is not a good/relevant number here.

 

[solderdude]

Well... 2nd harmonic at -60dB would be masked. Ofcourse the IM products that always accompany harmonic distortion, may not be masked and will not contribute to pleasantness but rather smearing.
Regardless of how much you want HD to be the only 'product' it isn't.

So what we can conclude is that you say it should not be too much distortion but it is not known when it is too much and it also is not clear what the minimal amount of distortion is.

Do you know of any research that shows what the minimum is ?
I mean ... no distortion is said to sound 'lifeless' so there must be some minimum (subjectively determined) where sound will 'improve' and as above a certain point it must become less 'nice' again.
There thus must be a sweet spot and that should be quite provable with listening tests.

In any case you can use the program distort and determine that for yourself. You can even choose which profile (H2 only or H2+H3 in a certain ration).
Maybe you can be a bit less unsecure about the values you mention.
I mean ... 20dB difference (in harmonics) is huge.
Not only the fundamental gets harmonics added but also the harmonics get their own harmonics and their own IM products with other tones of other instruments.

 

[Mulder]

Well... 2nd harmonic at -60dB would be masked. Ofcourse the IM products that always accompany harmonic distortion, may not be masked and will not contribute to pleasantness but rather smearing.
Regardless of how much you want HD to be the only 'product' it isn't.

So what we can conclude is that you say it should not be too much distortion but it is not known when it is too much and it also is not clear what the minimal amount of distortion is.

Do you know of any research that shows what the minimum is ?
I mean ... no distortion is said to sound 'lifeless' so there must be some minimum (subjectively determined) where sound will 'improve' and as above a certain point it must become less 'nice' again.
There thus must be a sweet spot and that should be quite provable with listening tests.

In any case you can use the program distort and determine that for yourself. You can even choose which profile (H2 only or H2+H3 in a certain ration).
Maybe you can be a bit less unsecure about the values you mention.
I mean ... 20dB difference (in harmonics) is huge.
Not only the fundamental gets harmonics added but also the harmonics get their own harmonics and their own IM products with other tones of other instruments.

The majority of all music played in HiFi-systems is pop and rock. Even among audiophiles. I've thought about this, because almost all electric instruments have an inherent distortion. Sometimes a very high one too. There is also no natural reference. Do you know how e.g. a grand piano sounds, or a trumpet, without amplification, then you also know when it doesn't sound quite right. I am thinking about whether distortion in vinyl, tube amplifiers, etc. is simply masked, i.e. cannot be distinguished or identified, because the music itself contains distortion. It may even "soften" the music which may be pleasant. In true acoustic music, this same softening may be perceived as a lack of definition, i.e. smearing.

 

[Galliardist]

The majority of all music played in HiFi-systems is pop and rock. Even among audiophiles. I've thought about this, because almost all electric instruments have an inherent distortion. Sometimes a very high one too. There is also no natural reference. Do you know how e.g. a grand piano sounds, or a trumpet, without amplification, then you also know when it doesn't sound quite right. I am thinking about whether distortion in vinyl, tube amplifiers, etc. is simply masked, i.e. cannot be distinguished or identified, because the music itself contains distortion. It may even "soften" the music which may be pleasant. In true acoustic music, this same softening may be perceived as a lack of definition, i.e. smearing.

There has to be signal in the relevant frequencies, and a reasonable amount of sound, before masking of some distortion and noise can happen. You can hear this with LP where more normal levels of music have a sudden quiet moment, and the noise comes out: not only ticks and pops but also any constant surface noise. With anything digital, it's most unlikely you'll hear noise even if the amplifier is quite noisy.

As it happens, I listen to a lot of solo classical guitar and lute music, and there is little there to mask problems. It's commonly claimed that classical guitar is "easy" to reproduce, but it isn't, because the noise isn't masked, and the sound of different forms of the instrument (what soundboard bracing is used, and the composition of the treble strings) can get obscured. Some lattice braced guitars (look it up) can have a soft sound on the top two strings that is further softened by second harmonic, and it can happen with less than is normally claimed as having to be there to be audible.

I've seen it claimed that violins can come over as a more "idealised" sound with high second and third harmonic. I can see how such a sound may be preferable to a warts and all presentation. I've heard that kind of effect on some older recordings, but not caused by a playback system alone that I can remember.

Full orchestra masks quite successfully, so I don't think it is distortion so much as a full range sound that does the masking.

 

[Sokel]

The majority of all music played in HiFi-systems is pop and rock.

I think there a good amount of jazz and a smaller amount of us listening to classical.Mask can be successful on one hand but on the other there's stuff like Swan Lake or Scheherazade or,or... where really silent parts of one instrument playing (and creating a beautiful depth) where noise can be evident and distortion has nothing to contribute (is perfect by it's own).

So...

 

[Galliardist]

here's how a good engineer looks when he discovers that distortion doesn't actually sound bad
(not my image, just a crosslink from Audioholics - Stop Chasing SINAD Distortion in Audio Amplifiers)

It raises the question "what is better sounding anyway?" Is this "better" sound the equivalent of rose tinted spectacles?

Sometimes I think of it like those Facebook photos with those filters applied. The person posting thinks it looks great, but all I do is wonder what they really look like. What is it they are adding? Ah, yes... distortion.

 

[Chr1]

Even top models on magazine covers uisually get distorted via Photoshop these days. Not as obvious as the amateur attempts with phone filters on people's Facebook/Instagram.
Again, you could argue that it comes down to the quality of the source material and the amount of distortion added...A little tweaking might be considered an improvement. Subjective obviously.

 

[Sokel]

Even top models on magazine covers uisually get distorted via Photoshop these days. Not as obvious as the amateur attempts with phone filters on people's Facebook/Instagram.
Again, you could argue that it comes down to the quality of the source material and the amount of distortion added...A little tweaking might be considered an improvement. Subjective obviously.

I wouldn't say distorted,it's more like processed,like what EQ does to audio for example but that is on the production side,imagine wearing colored glasses to watch them.

 

[Chr1]

I beg to differ. I would suggest that the photograph equivalent of tweaking EQ is no more than adjusting the histogram or colour levels. As soon as you start removing facial blemishes, whitening teeth and eyes etc, I reckon you are into the realm of distorting the image surely. Either way, you could argue that this is all part of the artistic process involved in photography these days. Whether we like it or not is subjective.

 

[Sokel]

Either way, you could argue that this is all part of the artistic process involved in photography these days. Whether we like it or not is subjective.

That's exactly what at said at the end of the post.
That's production,we don't care about it.

All we care is watch it with clean eyes.
The great confusion in ASR comes from people who actively involve in production and have a word on it so all the rest are carried away by it.
No,we only care about playback,production people get payed for that,we shouldn't waste valuable (unpaid) time on this.

 

[Chr1]

Again, this thread is purely about whether people like some types of distortion.
I have said several times previously that of course in playback it is not hi fidelity. We know this.

 

[antcollinet]

I beg to differ. I would suggest that the photograph equivalent of tweaking EQ is no more than adjusting the histogram or colour levels. As soon as you start removing facial blemishes, whitening teeth and eyes etc, I reckon you are into the realm of distorting the image surely. Either way, you could argue that this is all part of the artistic process involved in photography these days.

Exactly - just like the changes made at the mixing mastering stage are part of the creative process for music.

If you look at dictionary definitions of distortion, it is almost exclusively defined as a negative process: IE a process that changes something from it's natural or correct shape. Distortion of the truth etc etc.

I see this as the case also with audio. All the harmonics and unharmonic tones emmitted by an instrument are not distortion - they are the natural waveform of that instrument.

Similarly changes made during creation are not distortion - they are part of the creative processes to make the sound as desired by the artists.


Distortion is an unwanted process that subsequently distorts the shape of that music - altering the sound in ways not intended by the artist.


In particular in reproduction gear - in an uncontrolled fashion as a result of limitation is the linearity of the electronics. Not in any way desirable.

 

[pkane]

Nice graphs, thank you. Showing the math/measurement diffs is always a good start.
Problem is, pretty much the same happens when you add H2 to an instrument, to a track, to .. any sound, anywhere. And apparently it does (or at the very least can) sound good.

So .. you just need to prove that those graphs you are showing do actually sound 'bad'

In the meantime, here's one of the Gedlee papers Auditory Perception of Nonlinear Distortion.
They used/tested 21 transfer functions. Not in the papers but he seems to be ~active on diyaudio. Ask nicely, maybe you can get them (e.g. Erin managed to schedule an interview there).
Generally, that paper could/should have been more detailed but there are still a few clear things:

• the THD and IMD vs preference graphs (figure 2 & 3) show a clear (but admitedly slight) negative/inverse correlation: i.e. more THD/IMD kinda sounds 'better'. According to 37 blind testers.
• the relation between the THD & IMD numbers (table 1) seems to be .. 'complicated'. Seems to depend heavily on the transfer function. What is 100% clear though: more THD does not automatically mean more IMD.
• IIRC, an assertion like "THD never comes alone, more THD = more IMD" was posted as a fact/argument in this thread. Numbers show it to be quite wrong: some kinds of THD add loads of IMD, some add very little (maybe close to none!?).

Or am I reading anything wrong!?

P.S.
Pretty sure that I seen a statement somewhere about the output transformers used in SET/300B amps being very good at not generating much IMD. But it was a statement from Jensen (i.e. manufacturer). Does anyone have trafo HD/IMD measurements that we can check?

> So .. you just need to prove that those graphs you are showing do actually sound 'bad'

Do I? All I need to show is that harmonics produced by musical instruments are different than IMD produced by electronics. Whether you like the extra distortion or not is not my concern, although I've now built two separate tools for you to listen and evaluate this for yourself.

> IIRC, an assertion like "THD never comes alone, more THD = more IMD" was posted as a fact/argument in this thread. Numbers show it to be quite wrong: some kinds of THD add loads of IMD, some add very little (maybe close to none!?).

Your discussion of THD vs IMD is very wrong, as has already been pointed out. Stop arguing a point that has no basis in reality. THD does not cause IMD, THD is IMD, just tested with a single tone. THD and IMD are not something independent -- they ARE THE SAME THING, caused by the same non-linearity. To continue to argue that somehow they are separate and distinct is simply ignorant.

 

[Mulder]

There has to be signal in the relevant frequencies, and a reasonable amount of sound, before masking of some distortion and noise can happen. You can hear this with LP where more normal levels of music have a sudden quiet moment, and the noise comes out: not only ticks and pops but also any constant surface noise. With anything digital, it's most unlikely you'll hear noise even if the amplifier is quite noisy.

As it happens, I listen to a lot of solo classical guitar and lute music, and there is little there to mask problems. It's commonly claimed that classical guitar is "easy" to reproduce, but it isn't, because the noise isn't masked, and the sound of different forms of the instrument (what soundboard bracing is used, and the composition of the treble strings) can get obscured. Some lattice braced guitars (look it up) can have a soft sound on the top two strings that is further softened by second harmonic, and it can happen with less than is normally claimed as having to be there to be audible.

I've seen it claimed that violins can come over as a more "idealised" sound with high second and third harmonic. I can see how such a sound may be preferable to a warts and all presentation. I've heard that kind of effect on some older recordings, but not caused by a playback system alone that I can remember.

Full orchestra masks quite successfully, so I don't think it is distortion so much as a full range sound that does the masking.

I didn´t mean "masked" in the proper sense of the word, i.e. that distortion is hidden behind or beneath the signal. What I meant was that, even if you are able to hear the distortion, it is perhaps possible that you don´t hear it as distortion since it will became a inseparable part of the sound. (With "you" I refer to anybody - not YOU.) Noise is of course something else and another story.

 

[Chr1]

Exactly - just like the changes made at the mixing mastering stage are part of the creative process for music.

If you look at dictionary definitions of distortion, it is almost exclusively defined as a negative process: IE a process that changes something from it's natural or correct shape. Distortion of the truth etc etc.

I see this as the case also with audio. All the harmonics and unharmonic tones emmitted by an instrument are not distortion - they are the natural waveform of that instrument.

Similarly changes made during creation are not distortion - they are part of the creative processes to make the sound as desired by the artists.


Distortion is an unwanted process that subsequently distorts the shape of that music - altering the sound in ways not intended by the artist.


In particular in reproduction gear - in an uncontrolled fashion as a result of limitation is the linearity of the electronics. Not in any way desirable.

Kind of agree. However, what if the original photograph/recording is not particularly high quality?
If you wanted to reproduce it to put on your wall, is it acceptable to tweak it slightly in Photoshop first?
To me, this is the equivalent of tweaking the sound profile in playback.

 

[lashto]

It raises the question "what is better sounding anyway?" Is this "better" sound the equivalent of rose tinted spectacles?

Sometimes I think of it like those Facebook photos with those filters applied. The person posting thinks it looks great, but all I do is wonder what they really look like. What is it they are adding? Ah, yes... distortion.

'better' is in the eye of the beholder ... may very well be rosa tinted

Just some thoughts from my experience with other people' preferences..

Many/most EEs seem to prefer the distortion "free" sound. Seems to be an educated-preference. Could it be that most (all?) people will develop the same taste if they had the same information/experience? Maybe ... or may be not: e.g. it did not really work for me. I like some 120SINAD devices but also still like tubes/turntables and often prefer them.

Talking about the 7 billion people-at-large, here's my 'statistic':
...vast majority of them do not even know the word "distortion"
...vast majority of the ones who (at least) heard it, couldn't care less
...and a large part of that tiny minority who ~knows and does care, still keeps buying millions of tubes, turntables, records, distort pedals/plugins/effects/etc..
AFAICS, the people who know about distortion, care about it and actually prefer distortion-"free" are a minority. A seriously tiny one!

But are they actually right? Maybe .. maybe not. Hard to say unless we have a whole record/playback chain above 120SINAD (including mics, speakers.. everything).

However, we can clearly say that people crave distortion (at the very least, the one already embedded in the music). And they also quite crave the 'extra' distortion added by studios. For me the 'safe' hypothesis is that they'll continue to like (at least) those.

 

[UliBru]

At the production stage there is someone who makes a decision: the production is ready, the product will be deliviered as it is.
Thus the product arrives at the reproduction stage.
Now if the product is ok it should be used 1:1 without any further "sounding". Raising or lowering the playback level is possibly the only fully transparent change (thinking about loudness curves a different level will also change the timbre)..

BUT:
the decision of the producer is based on what he hears. He is using his own equipment and environment.
At the reproduction side the equipment is different and may even sound different despite of the theoretical 1:1 transfer.
So it is possible that the listener likes to have some changes, a bit here and a bit there. This could be equalizing or also adding distortion.
We may call it sounding. The problem with sounding: usually the listener has less control, the sounding is changing the complete track.
An electric guitar can get heavy distortions at the production side, a voice is kept untreated. Applying the same distortion at the reproduction side will also influence the voice.

A 2nd harmonic distortion is assumed to add frequencies one octave higher. The typical listener expects that this sounds harmonious. Like a piano melody played with a 2nd melody one octave higher. But this is not the case. The 2nd harmonic distortion also adds intermodulation distortions. You can test it: generate a multitone signal, add a 2nd HD and see the new frequency response.

And finally I recommend to create a track with 2nd HD by one of the available tools, to subtract the original track and then to listen to the boosted pure 2nd HD. Then you know what is added in reality. The discussion on how much of the distortion is acceptable is just a dicussion about how much is perceived or masked.