Do we crave distortion?

[DVDdoug]

There is a difference between production and reproduction. Guitar amplifiers are designed to distort/saturate in a "pleasing way" when driven hard. But you wouldn't want the to play your music through a guitar amp. Similarly, "exciter" effects are typically used during production.

A "high fidelity" reproduction system is supposed to accurately reproduce the recording. Or at least be capable of accuracy. The listener may want to "enhance" the sound.

P.S.
Dan Clark says headphones with distortion are sometimes described as having "more detail". And he says something about bass distortion giving the impression of stronger bass.

 

[Zensō]

While the results of your anecdotal test are very interesting, I personally wouldn’t want to fuss with changing distortion levels throughout the day as I switch from one type of music to another. With that in mind, I opt for the most distortion-free signal path possible.

 

[Galliardist]

Well, yeah. I didn't think I needed to state that. But of course it's "the copy of the recording" then. CD / Streaming.

I know I was being pedantic. But the difference seems to raise questions in the subject of this thread. Do we really want to accurately reproduce a bad recording if help can be at hand? Should we resist if, say, sometime in the future, AI or some algorithm could do smart things to retrieve such bad recordings without destroying the musical information?

 

[Deleted member 21219]

While the results of your anecdotal test are very interesting, I personally wouldn’t want to fuss with changing distortion levels throughout the day as I switch from one type of music to another. With that in mind, I opt for the most distortion-free signal path possible.

I agree. IMO, this is listener-dependent. Some people like to fuss and fiddle, some don't. Some people are tolerant of poor recordings, and others less so.

Personally, I am rather tolerant of bad recordings, but very intolerant of bad performances, and I have never found euphony to improve a performance. YMMV.

Jim Taylor

 

[Rednaxela]

Maybe a lot of music, esp. the older stuff, was made with and for the presence of distortion in the playback chain?

What I mean is when all you had was tubes, both in the studio and in the living room, why would a recording from that era sound nicer on 100% transparent gear?

 

[Keith_W]

There is a difference between production and reproduction.

Yes, obviously I know that. I am not interested in philosophical discussions about what "high fidelity" means, because it means "fidelity to the recording" and that is that. I said so in my first post. What I am interested in, is whether some added distortion actually subjectively enhances the sound, not a dogmatic assertion that as little distortion as possible sounds the best. This conflates "better measuring" with "better sounding", and I do not think that better measuring necessarily means better sounding. The abundance of audiophiles who prefer poor measuring systems is enough preliminary evidence that this phenomenon should be investigated. Yes, in the end it could all turn out to be some kind of cognitive bias, but we don't know until we investigate, do we? Maybe (as the title says) we crave distortion?

I have performed my own single blind test on at least 10 subjects (I have lost count), which shows a significant preference towards some added distortion. What I want to know is if there has been a more controlled study which affirms/refutes it, and whether people have tried it. Otherwise, I can see a potential AES publication if this has not been extensively studied before (apparently it has, thanks for the Geddes-Lee links @MarnixM - I will go and read them).

 

[Zensō]

Yes, obviously I know that. I am not interested in philosophical discussions about what "high fidelity" means, because it means "fidelity to the recording" and that is that. I said so in my first post. What I am interested in, is whether some added distortion actually subjectively enhances the sound, not a dogmatic assertion that as little distortion as possible sounds the best. This conflates "better measuring" with "better sounding", and I do not think that better measuring necessarily means better sounding. The abundance of audiophiles who prefer poor measuring systems is enough preliminary evidence that this phenomenon should be investigated. Yes, in the end it could all turn out to be some kind of cognitive bias, but we don't know until we investigate, do we? Maybe (as the title says) we crave distortion?

I have performed my own single blind test on at least 10 subjects (I have lost count), which shows a significant preference towards some added distortion. What I want to know is if there has been a more controlled study which affirms/refutes it, and whether people have tried it. Otherwise, I can see a potential AES publication if this has not been extensively studied before (apparently it has, thanks for the Geddes-Lee links @MarnixM - I will go and read them).

One thing I'm curious about, were the distorted and undistorted signals volume matched? If not, there's a possibility the distortion plugin increased the signal slightly, which we all know could also explain the preference for the distorted signals. Not saying this is the case one way or the other, but to truly suss out what is happening would require level matching.

 

[Keith_W]

One thing I'm curious about, were the distorted and undistorted signals volume matched? If not, there's a possibility the distortion plugin increased the signal slightly, which we all know could also explain the preference for the distorted signals. Not saying this is the case one way or the other, but to truly suss out what is happening would require level matching.

Good question! I suspected there was a level difference, so I asked @pkane. Answer is here.

 

[krabapple]

For some time now, I have been running @pkane's PKHarmonic VST which has the ability to add various amounts of harmonic distortion. I have also played with some Exciter VST plugins which (unlike PK) allow for targeted addition of harmonic distortion - for example, you can add high frequency harmonic spectra and produce the illusion of clarity, or low frequency spectra which makes playback sound fuller.

A number of people have visited me and listened to my system. They do not know that I am adding harmonic distortion to my system. I simply ask them if they prefer A or B (single blind test).

Do you mix up A and B, or ,e.g. is A always clean and B always distorted?

 

[krabapple]

Harmonics of a single cleanly played note on an electric bass guitar, electrical signal into an ADC, no added distortions.

It took several tries to pluck so that the 2nd harmonic was not at a higher level than the fundamental.

Is that a G? Open string, or fretted?

 

[solderdude]

I knew there was a scientific reason to like beer!

Talking about beer...

What happens if you null a song with the same song run through distort with the settings as described ?

You can consume a beer while doing this which sadly is the only reference to beer.
Well... 'beer' is Dutch for 'bear' and beer is called 'bier' in the Netherlands.

0.2% 2nd harmonic should not alter tonality much, let alone 0.004% 3rd harmonic.

 

[Keith_W]

Do you mix up A and B, or ,e.g. is A always clean and B always distorted?

When people come to my house, they all know that the system is DSP'ed to the hilt. I tell them that they will be comparing different DSP settings. I have also performed the same A-B experiment with different target curves, different crossover types, different crossover points, different room correction philosophies, whether I used MSO or not, uBACCH or no uBACCH, and various VST's. The intervention that has produced the strongest and most consistent preference effect is PKHarmonic, where there is unanimous agreement (from sample size of about 10) that some harmonic distortion sounds better. The effect is (surprisingly to me) even stronger than uBACCH, where I have had two dissenters prefer non-uBACCH. I never tell them what I am doing - I tell them that I am going to change a DSP setting, and ask them to indicate a preference. I let them switch back and forth as many times as they like, and I ask if they want it demonstrated with another piece of music. Then I collect the vote and tell them what the intervention was. At no time do they know what I have done, and I try my best not to indicate to them what my own preference is.

 

[LightninBoy]

As an amateur music producer, let me add this perspective. Many mastering plugins (among many other plugin types) provide a THD control to add THD to the final recording. I find using it makes a big difference in making the final recording sound similar to pro commercial recordings of popular music.

My theory on this: Music producers often talk about the importance of "gluing" the instruments together. I think this comes from the typical practice of recording individual instruments to their own tracks, then mixing them together to model a single performance. Without some processing, simply mixing them together doesn't sound ... right. And that's where this "glue" concept comes in: how does one blend these separately recorded instruments to make them sound like they weren't separately recorded. The trick is to send all the tracks through some type of shared processing. And typically there is an analog compressor or limiter in the final mixing bus which was doing compression, yes, but is also adding THD (initially as a side effect, but then eventually as an intentional addition). I think *both* of these contribute to the overall effect of the instruments blending together in a more believable way.

As consumers of commercial music, many of us have become accustomed to this sound. So a "pure" recording that doesn't add THD may sound strange to us. Total speculation here, but maybe the classical music examples you are using have far less compression/THD in the recording, so adding a little more makes it sound more "normal" to us. This may also explain why it doesn't improve on the metal music examples, which would already be compressed to hell and heavy doses of THD added to thicken the sound.

 

[Lieglein]

As an amateur music producer, let me add this perspective. Many mastering plugins (among many other plugin types) provide a THD control to add THD to the final recording. I find using it makes a big difference in making the final recording sound similar to pro commercial recordings of popular music.

My theory on this: Music producers often talk about the importance of "gluing" the instruments together. I think this comes from the typical practice of recording individual instruments to their own tracks, then mixing them together to model a single performance. Without some processing, simply mixing them together doesn't sound ... right. And that's where this "glue" concept comes in: how does one blend these separately recorded instruments to make them sound like they weren't separately recorded. The trick is to send all the tracks through some type of shared processing. And typically there is an analog compressor or limiter in the final mixing bus which was doing compression, yes, but is also adding THD (initially as a side effect, but then eventually as an intentional addition). I think *both* of these contribute to the overall effect of the instruments blending together in a more believable way.

As consumers of commercial music, many of us have become accustomed to this sound. So a "pure" recording that doesn't add THD may sound strange to us. Total speculation here, but maybe the classical music examples you are using have far less compression/THD in the recording, so adding a little more makes it sound more "normal" to us. This may also explain why it doesn't improve on the metal music examples, which would already be compressed to hell and heavy doses of THD added to thicken the sound.

There is nothing ebjective in all of this. That people may "crave" for distortion or a dense frequency spectrum is if at all a developement of the music industry and habits.
We just have to take a look back in time for example the 70's - 80's and we'll clearly see, that rock/metal music was not nearly as distorted and "glued together" as it is nowadays. People enjoyed it as well.

The thing is that "Do we crave for distortion?" is a question on a extremely fundamental level. "Is it just given?". And I highly doubt that if we take a look into the past. It has gotten more exciting maybe (because that's what an exciter or distortion does of course), but I think a very important factor as well is "niveau".

 

[Curvature]

When people come to my house, they all know that the system is DSP'ed to the hilt. I tell them that they will be comparing different DSP settings. I have also performed the same A-B experiment with different target curves, different crossover types, different crossover points, different room correction philosophies, whether I used MSO or not, uBACCH or no uBACCH, and various VST's. The intervention that has produced the strongest and most consistent preference effect is PKHarmonic, where there is unanimous agreement (from sample size of about 10) that some harmonic distortion sounds better. The effect is (surprisingly to me) even stronger than uBACCH, where I have had two dissenters prefer non-uBACCH. I never tell them what I am doing - I tell them that I am going to change a DSP setting, and ask them to indicate a preference. I let them switch back and forth as many times as they like, and I ask if they want it demonstrated with another piece of music. Then I collect the vote and tell them what the intervention was. At no time do they know what I have done, and I try my best not to indicate to them what my own preference is.

Please see my note about (psychoacoustic) loudness. You need to adjust the output of the PKHarmonic'ed signal such that it's slightly lower, otherwise what your listeners are reacting to is the same thing everyone else does in listening tests without level control: there are always qualitative comments about the difference in sound simply to do with A being louder than B or vice versa.

This is a well known factor in subjective testing.

Edit: SPL or voltage is physical. Loudness is the sensation. A louder sound has more energy than a quieter sound, and that energy can come from nonlinear distortion or from higher output, and can be compensated by offsetting one or the other. You've controlled for one factor (output, assuming you've been careful about the signal chain), not the other (subjective loudness). I would bet that the qualitative comments will no longer have place once the level adjustment is made.

 

[Rednaxela]

And careful not to do this too much.

 

[Zensō]

Edit: SPL or voltage is physical. Loudness is the sensation. A louder sound has more energy than a quieter sound, and that energy can come from nonlinear distortion or from higher output, and can be compensated by offsetting one or the other. You've controlled for one factor (output, assuming you've been careful about the signal chain), not the other (subjective loudness). I would bet that the qualitative comments will no longer have place once the level adjustment is made.

Agreed. It would be interesting to continue these tests, but with the distorted signal slightly audibly quieter to see if the results flip.

 

[fpitas]

Maybe a lot of music, esp. the older stuff, was made with and for the presence of distortion in the playback chain?

What I mean is when all you had was tubes, both in the studio and in the living room, why would a recording from that era sound nicer on 100% transparent gear?

I've entertained this. Not so much that they didn't care about accurate gear, but that the reality was they didn't usually have it. So, they made the recording sound good with what they had.

 

[pkane]

Talking about beer...

What happens if you null a song with the same song run through distort with the settings as described ?

You can consume a beer while doing this which sadly is the only reference to beer.
Well... 'beer' is Dutch for 'bear' and beer is called 'bier' in the Netherlands.

0.2% 2nd harmonic should not alter tonality much, let alone 0.004% 3rd harmonic.

I did some PKH vs no PKH analysis in another thread, using about -45dB THD settings in PKHarmonic.

Nulled against the original file (remastered Beatles track). Harmonics were set a sfollows:

PK Metric:

Error distribution (sample by sample):

Phase:

Just a very zoomed-in comparison of the waveforms (white is with the PKH harmonics added):

 

[solderdude]

The one from @Keith_W has even less difference in the files.
At least we know the frequency response did not change (and was unlikely to begin with)

The white waveform seems to have gotten a small offset, rather than different amplitude for some reason.