Distortion as an effect - but with dynamics?

[RobC]

Clearly, if you have a good, dynamic instrument, you don't want to trash its signal. But that's what distortion, even as an effect, does.

Right away, the best solution seems to do some parallel, dry/wet kind of processing. Mixing the clean sound with the distortion, would sound as ugly as parallel compression, but still better than affecting the whole signal.

One possibility might be converting the original signal to CV, and make it control the distortion's audio leveling / dynamics. That way the sound would have the characteristics (harmonics?), while remaining dynamic. The more squashed the sound, the more leveling the CV should do - gladly, scaling (EDIT: amount should do) can be controlled, so that with a dry/wet mix sounds flexible in my imagination.

Question is, how accurately does Thor follow waveforms when it comes to converting audio to CV? Or are there better options for that conversion?

(By the time I finish the question, I seem to get the answer. Some new level retardation that is... xD)

[Loque]

What i do, if i want some additional harmonics, but do not want a completely destroyed audio signal, i use a parallel channel with a EQ before the distortion/saturation/whatever. Mostly some mid frequency part (BP filter) that goes through the amp fx and i mix it back.

And sometimes i use the technique you describe and want to achieve to lower the distortion. I use a compressor and its gain reduction CV output to control additional parameters, like EQs or in your case the distortion amount. Works quite well, but you should keep in mind, that you have a delay. With some tricks it might be compensated. I would use this technique instead of audio to cv, because a audio signal has lots of fluctuations in it's amplitudes, where the compressor smooths this out and you can also smooth the generated CV gain reduction with attack/release of the compressor and scale it with Threshold or Ratio.

To go a step further, you can EQ the signal BEFORE the compressor to handle different frequencies more or less agressive.

Here is a simple setup:

[Timmy Crowne]

This is a really good question! Loque is right that an envelope follower might be better for your concept than audio-to-cv.

Audio waveforms are usually comprised of positive and negative phases. Using such a signal to modulate the gain of your source would be amplitude modulation and would produce a lot of sidebands that aren’t harmonic to the original tone.

A fast envelope follower would likely do a better job, though you might get the cleanest results from bouncing to audio and manually editing the amplitudes to be more similar to the original. This is all theoretical though because I’m at work right now!

[Marco Raaphorst]

saturation is dynamic. it responds to the signal in a dynamic way, non-lineair.

so I don't get what's not dynamic about it. I am a guitar player and when using good saturation devices I can adjust the amount of saturation by my playing dynamics.

[normen]

saturation is dynamic. it responds to the signal in a dynamic way, non-lineair.

so I don't get what's not dynamic about it. I am a guitar player and when using good saturation devices I can adjust the amount of saturation by my playing dynamics.

Yeah but the dynamics are gone - transformed into distortion ^^

About the question of OP: Did you try Seligs coloring EQ? Putting distortion where it's effective and leaving it out where theres a lot of dynamics might be an option?

[O1B]

You don't want Distortion.

you don't want to trash its signal.

For just a lil' Cream in your Coffee, You want Drive.

If you insist on using Distortion, better it be on with a separate NAKED and DRESSED pot - instead of your standard DRY/WET:


Pulverizer can do these duties. Squash n Dirt. SCREAM, too (parallel)
But, sometimes, you just need a faster Env than on Scream or Pulverizer.

... what to do...

Clearly, if you have a good, dynamic instrument, you don't want to trash its signal. But that's what distortion, even as an effect, does.

Right away, the best solution seems to do some parallel, dry/wet kind of processing. Mixing the clean sound with the distortion, would sound as ugly as parallel compression, but still better than affecting the whole signal.

One possibility might be converting the original signal to CV, and make it control the distortion's audio leveling / dynamics. That way the sound would have the characteristics (harmonics?), while remaining dynamic. The more squashed the sound, the more leveling the CV should do - gladly, scaling (EDIT: amount should do) can be controlled, so that with a dry/wet mix sounds flexible in my imagination.

Question is, how accurately does Thor follow waveforms when it comes to converting audio to CV? Or are there better options for that conversion?

(By the time I finish the question, I seem to get the answer. Some new level retardation that is... xD)

[Timmy Crowne]

So I did an experiment:

• Imported and normalized raw kick and snare samples. (Seen in green.)

• Ran original samples through Saturation Knob, cranked to max, and bounced the output. (Seen in red. @Marco, I think this is what OP is talking about. The dynamics of the original sample are completely flattened out. Of course, the distortion setting used here is extreme, but I wanted to see what was possible.)

• Automated the level of the new distorted samples to approximate the envelopes of their original sources and bounced the output. (Seen in purple. I also high-passed the distorted kick sample at 30hz because the distortion was imposing a nasty DC offset.)

Screen Shot 2018-04-04 at 5.22.41 PM.png (681.11 KiB) Viewed 3062 times

.
Here's a short drum loop using the samples in the test. 2 bars of original, 2 bars of distorted, 2 bars of re-enveloped after distortion:



All three instances sound very different from each other, with the final re-enveloped samples having the brightness and crunch imparted by the distortion while retaining the overall dynamic shape of the original samples. It'll take a better mind than mine to figure out how to do this in realtime, though...

[avasopht]

Okay, .. I actually made an effect that adds distortion while preserving dynamics. It was an unreleased Rack Extension I made back in 2014.

Basically it calculates the RMS, then compresses the signal above a threshold relative to the RMS. There is also a function to apply up to 80ms of lookahead.

Although it was a dynamics effect, when pushed produced distortion that could be used creatively.

Here is a demo of it being abused to produce a little distortion:

[Marco Raaphorst]

saturation is dynamic. it responds to the signal in a dynamic way, non-lineair.

so I don't get what's not dynamic about it. I am a guitar player and when using good saturation devices I can adjust the amount of saturation by my playing dynamics.

Yeah but the dynamics are gone - transformed into distortion ^^

About the question of OP: Did you try Seligs coloring EQ? Putting distortion where it's effective and leaving it out where theres a lot of dynamics might be an option?

Saturations adds to rms value because the signal gets more complicated because it includes more tones, more complexity. Would not call this lack of dynamics. I would call it a more colorfull tone.

Check the sound of Stevie Ray Vaughn for example. Super lively and dynamic, because of superb use of saturation. A clean sound can not have that dynamic impact. But that break up adds to the dymamics.

It not “saturation kills dynamics” it is “over using saturation will ruin sound”. But saturation is a dynamic effect. A coloring tool like Selig also noticed

[vectro]

Automated the level of the new distorted samples to approximate the envelopes of their original sources and bounced the output.

Use the vocoder as a multiband envelope follower and connect cv from the most suitable band to control fader for the distorted channel.

[normen]

Yeah but the dynamics are gone - transformed into distortion ^^

About the question of OP: Did you try Seligs coloring EQ? Putting distortion where it's effective and leaving it out where theres a lot of dynamics might be an option?

Saturations adds to rms value because the signal gets more complicated because it includes more tones, more complexity. Would not call this lack of dynamics. I would call it a more colorfull tone.

Check the sound of Stevie Ray Vaughn for example. Super lively and dynamic, because of superb use of saturation. A clean sound can not have that dynamic impact. But that break up adds to the dymamics.

It not “saturation kills dynamics” it is “over using saturation will ruin sound”. But saturation is a dynamic effect. A coloring tool like Selig also noticed

Well I don‘t know. Limiting with a ratio of infinity to one and zero attack and release is what I call reducing dynamics - and thats what a distorting amp does. But maybe you‘re using that terminology differently.

[Marco Raaphorst]

Saturations adds to rms value because the signal gets more complicated because it includes more tones, more complexity. Would not call this lack of dynamics. I would call it a more colorfull tone.

Check the sound of Stevie Ray Vaughn for example. Super lively and dynamic, because of superb use of saturation. A clean sound can not have that dynamic impact. But that break up adds to the dymamics.

It not “saturation kills dynamics” it is “over using saturation will ruin sound”. But saturation is a dynamic effect. A coloring tool like Selig also noticed

Well I don‘t know. Limiting with a ratio of infinity to one and zero attack and release is what I call reducing dynamics - and thats what a distorting amp does. But maybe you‘re using that terminology differently.

You are right. But weird thing is that I don't hear it as compression. And even using a compressor often results in a more lively and dynamic tone to my ears. But I agree, that's the wrong terminology.

In the end it only matter how it sounds. Not how we call it. Best stuff we cannot explain. That magic Stevie Ray Vaugh tone? We can't explain it.

[normen]

In the end it only matter how it sounds. Not how we call it.

Well thats true for the moment / production / performance but not if we want to talk about it right? Agreed on everything else.

[Timmy Crowne]

Use the vocoder as a multiband envelope follower and connect cv from the most suitable band to control fader for the distorted channel.

Vectro this is a great idea. Might be just the ticket to get a nice crunchy sound, while still retaining dynamics. I’ll have to try it later.

[selig]

Well I don‘t know. Limiting with a ratio of infinity to one and zero attack and release is what I call reducing dynamics - and thats what a distorting amp does. But maybe you‘re using that terminology differently.

You are right. But weird thing is that I don't hear it as compression. And even using a compressor often results in a more lively and dynamic tone to my ears. But I agree, that's the wrong terminology.

In the end it only matter how it sounds. Not how we call it. Best stuff we cannot explain. That magic Stevie Ray Vaugh tone? We can't explain it.

An extreme example - on a guitar amp, distortion adds harmonics, AND adds sustain. Compressors also add sustain using the same technique: reduce dynamics. A blues guitarist typically uses less 'drive', thus less compression, but it's still the same thing. And just like two compressors don't always sound the same, two amps don't always sound the same. And two guitarists definitely don't sound the same…

But they are all doing the same thing at the dynamics level: clipping the signal in some way (which is adding sustain), and because the waveform is being warped/distorted, also adding harmonics. The difference is with compressors we try to restrict adding harmonics to as little as possible, and with amps we try to enhance adding harmonics as much as possible.

Saturation or distortion at any level reduces dynamics, it's just that in many cases it's not reducing dynamics very much so you don't hear it as "compression". It could also be the adding of harmonics "clouds" the listener's ability to also hear the added sustain, except for in extreme cases.

Here's the first 100ms or so of a guitar note (1972 Fender Tele, if anyone's asking).
The first is the direct guitar, the second is Kussa Cream "Clean 1", the third is Kussa Vermillion Bluesy.
You can see how the initial transient is "limited" (reduced) even at these very clean settings. The end result is reduced dynamics.

Screen Shot 2018-04-05 at 9.51.13 AM.png (194.8 KiB) Viewed 2965 times

Taken to an extreme, here are the same three but using "Heavy Amp/overdrive" and "Too Big" and zoomed out to show how much compression the amps are adding:

Screen Shot 2018-04-05 at 9.55.46 AM.png (89.3 KiB) Viewed 2965 times

All examples are aligned to peak at the same level (-12dBFS) for comparison purposes.

[Marco Raaphorst]

You are right. But weird thing is that I don't hear it as compression. And even using a compressor often results in a more lively and dynamic tone to my ears. But I agree, that's the wrong terminology.

In the end it only matter how it sounds. Not how we call it. Best stuff we cannot explain. That magic Stevie Ray Vaugh tone? We can't explain it.

An extreme example - on a guitar amp, distortion adds harmonics, AND adds sustain. Compressors also add sustain using the same technique: reduce dynamics. A blues guitarist typically uses less 'drive', thus less compression, but it's still the same thing. And just like two compressors don't always sound the same, two amps don't always sound the same. And two guitarists definitely don't sound the same…

But they are all doing the same thing at the dynamics level: clipping the signal in some way (which is adding sustain), and because the waveform is being warped/distorted, also adding harmonics. The difference is with compressors we try to restrict adding harmonics to as little as possible, and with amps we try to enhance adding harmonics as much as possible.

Saturation or distortion at any level reduces dynamics, it's just that in many cases it's not reducing dynamics very much so you don't hear it as "compression". It could also be the adding of harmonics "clouds" the listener's ability to also hear the added sustain, except for in extreme cases.

Here's the first 100ms or so of a guitar note (1972 Fender Tele, if anyone's asking).
The first is the direct guitar, the second is Kussa Cream "Clean 1", the third is Kussa Vermillion Bluesy.
You can see how the initial transient is "limited" (reduced) even at these very clean settings. The end result is reduced dynamics.
Screen Shot 2018-04-05 at 9.51.13 AM.png

Taken to an extreme, here are the same three but using "Heavy Amp/overdrive" and "Too Big" and zoomed out to show how much compression the amps are adding:
Screen Shot 2018-04-05 at 9.55.46 AM.png

All examples are aligned to peak at the same level (-12dBFS) for comparison purposes.

Yes, I understand. Saturation creates a compressed signal.

But it also adds something funky. The inverted saturation with phase invered (tchad blake) trick adds a lot of oompf to the bassdrums for example. That's added energy to my ears. Don't need to understand why, but I am hearing a big tone. Like the tube amps. And that's compression + overtones I guess.

Anyways... back to playing my Tele...

[selig]

As for the actual process of adding the original dynamics back into a distorted signal, it's not simple since you need to calibrate all levels to be sure you're doing what you think you're doing.

Extracting the original dynamics is the first part. Ideally what we need is a simple peak detector, with no threshold and no attack. Release time would need to be fairly fast to track most waveforms, but no one setting will be perfect for all sources in my experience, even when using auto-release techniques. And the CV output of this device would ideally be calibrated in dB…

You'll also need a VCA type device that allows CV control of gain, preferably also calibrated in dB for easy setup with the peak detector.

So far we would have a peak detector that would take the input levels, convert them to DC (rectify) and then to dB, such that an input of 0 dBFS would add no gain, and levels below that would subtract gain (input of -10 dBFS subtracts 10 dB gain. But the tricky part is this will only work if the loudest input level was 0 dBFS. If it's lower, say -12 dBFS, you need to calibrate so that an input of -12 dBFS gives you an output CV of 0 dB (no change), but that's only part one of the calibration process…

Then you'll need to calibrate the output of your distortion device to the same known level, so that when you apply your dynamics to this signal the original levels will be roughly restored. Again for the sake of example we'll say your highest input peak hits -12dBFS, so you'll want to calibrate the output of your distortion device to also hit this level at the highest input peak.

OK, so what we end up with is signals that match the highest level, in our example -12 dBFS, are left alone, since they are already where you want them. A lower signal will likely be raised by the distortion effect in relation to the highest (see the extreme examples in my previous post), so it's THESE signals you'll want to turn down to remove all compression effects from the distorted signal.

What you'll end up with is basically an expansion effect, where the highest signals are left alone and the lower signals are reduced further - but ideally reduces EXACTLY how much they need to be reduced in order to match the input. This is the tricky part, because for each amount of drive in your distortion/saturation effect, you'll need to re-callibrate a different output curve/response to "undo" the compression. The only way 100% of the gain change effect would give expected results is IF you squashed your signal so ALL dynamics were eliminated down to the lowest level you want to represent. But that's unlikely and so it gets tricky.

Say for example a signal 24 dB below the highest peak gets increased by 12 dB (a 2:1 ratio), you'll need a 1:2 expansion ratio to restore the original dynamics - and this assumes this ratio is constant at all level, which it is likely not. But you can come close by calibrating to two points: the highest peak, and a lower level of your choosing (-24 is not a bad choice IMO).

Again, even after going through all of this, it will only work while your distortion effect is giving you 2:1 - if you change the amount of drive etc you'll need to re-calibrate your system!

It is for these reasons I'd approach this as a creative effect rather than a "precision" effect, since it will be difficult to exactly restore the original dynamics in all cases.

[selig]

Yes, I understand. Saturation creates a compressed signal.

But it also adds something funky. The inverted saturation with phase invered (tchad blake) trick adds a lot of oompf to the bassdrums for example. That's added energy to my ears. Don't need to understand why, but I am hearing a big tone. Like the tube amps. And that's compression + overtones I guess.

Anyways... back to playing my Tele...

Yes, compression can add energy, that's for sure! Throw in additional harmonics and you're adding energy in two dimensions/domains: Time (compression) and Frequency (harmonics).

I was responding to your comment that you would not call saturation "lack of dynamics". We are back on the same page then that saturation IS lack of dynamics, aka a "compressed signal"?

What I would say instead is that distortion/saturation is a non-linear (with regards to level) effect. It responds to dynamics, which may be what you were trying to say initially? The louder/harder you play, the more it saturates/distorts, which is a "dynamic" effect, but it is also reducing the total dynamics of the original signal while it does this. Maybe that's where any confusion of terms has entered this discussion.

BTW, as you add harmonics, even though the level stays the same, the loudness increases because of extra energy in the frequency domain.

[Loque]

As for the actual process of adding the original dynamics back into a distorted signal, it's not simple since you need to calibrate all levels to be sure you're doing what you think you're doing.

Extracting the original dynamics is the first part. Ideally what we need is a simple peak detector, with no threshold and no attack. Release time would need to be fairly fast to track most waveforms, but no one setting will be perfect for all sources in my experience, even when using auto-release techniques. And the CV output of this device would ideally be calibrated in dB…

You'll also need a VCA type device that allows CV control of gain, preferably also calibrated in dB for easy setup with the peak detector.

So far we would have a peak detector that would take the input levels, convert them to DC (rectify) and then to dB, such that an input of 0 dBFS would add no gain, and levels below that would subtract gain (input of -10 dBFS subtracts 10 dB gain. But the tricky part is this will only work if the loudest input level was 0 dBFS. If it's lower, say -12 dBFS, you need to calibrate so that an input of -12 dBFS gives you an output CV of 0 dB (no change), but that's only part one of the calibration process…

Then you'll need to calibrate the output of your distortion device to the same known level, so that when you apply your dynamics to this signal the original levels will be roughly restored. Again for the sake of example we'll say your highest input peak hits -12dBFS, so you'll want to calibrate the output of your distortion device to also hit this level at the highest input peak.

OK, so what we end up with is signals that match the highest level, in our example -12 dBFS, are left alone, since they are already where you want them. A lower signal will likely be raised by the distortion effect in relation to the highest (see the extreme examples in my previous post), so it's THESE signals you'll want to turn down to remove all compression effects from the distorted signal.

What you'll end up with is basically an expansion effect, where the highest signals are left alone and the lower signals are reduced further - but ideally reduces EXACTLY how much they need to be reduced in order to match the input. This is the tricky part, because for each amount of drive in your distortion/saturation effect, you'll need to re-callibrate a different output curve/response to "undo" the compression. The only way 100% of the gain change effect would give expected results is IF you squashed your signal so ALL dynamics were eliminated down to the lowest level you want to represent. But that's unlikely and so it gets tricky.

Say for example a signal 24 dB below the highest peak gets increased by 12 dB (a 2:1 ratio), you'll need a 1:2 expansion ratio to restore the original dynamics - and this assumes this ratio is constant at all level, which it is likely not. But you can come close by calibrating to two points: the highest peak, and a lower level of your choosing (-24 is not a bad choice IMO).

Again, even after going through all of this, it will only work while your distortion effect is giving you 2:1 - if you change the amount of drive etc you'll need to re-calibrate your system!

It is for these reasons I'd approach this as a creative effect rather than a "precision" effect, since it will be difficult to exactly restore the original dynamics in all cases.

Sounds interesting, but also very (maybe too) complicated. Nevertheless, if there is a high dynamic in the audio, the results may not be satisfying at all and manual adjustments are reqeuired.

What about adding some expension/compression/limiting to the audio scanner that controls the distortion? and after mixing the original and distorted signal back together, their dynamic can be touched by a compressor/expander with the "scanned" amount. This should adjust the amount of distortion in a specific range only and also adjusts the final audio signal based on the changes made to the audio signal. Of course, it requires always some adjustment for different audio material and would not be 100% accurate in db without perfect measurement. But i guess you can get quite close. What do you think?

[selig]

Sounds interesting, but also very (maybe too) complicated. Nevertheless, if there is a high dynamic in the audio, the results may not be satisfying at all and manual adjustments are reqeuired.

What about adding some expension/compression/limiting to the audio scanner that controls the distortion? and after mixing the original and distorted signal back together, their dynamic can be touched by a compressor/expander with the "scanned" amount. This should adjust the amount of distortion in a specific range only and also adjusts the final audio signal based on the changes made to the audio signal. Of course, it requires always some adjustment for different audio material and would not be 100% accurate in db without perfect measurement. But i guess you can get quite close. What do you think?

The simplest would be to side-chain an expander on the output of the effect. The side chain input would be the ‘raw’ signal coming into the effect. This will still require some tweaking, and the results won’t be perfect - but maybe that’s not required for this.

In fact, I’m not sure it’s required at all. When I’ve done similar things in the past I’ve not been as satisfied with the results as I expected. To my ears it sounded more like gating the output rather than restoring the dynamics. Thing is, saturation, while “limiting” the signal also adds harmonics, which in essence retain the dynamics. Louder signals create more harmonics, which sounds louder (even though the peak level is restrained).

The only advantage to this approach IMO would be to remove some of the sustain that saturation/distortion adds - but in most cases that’s part of the charm!


Try the side-chain expander idea and see if that gives useful results. But before that I’d be sure about what I was hearing in the signal that I didn’t like, and make sure it IS a lack of dynamics that’s bothering you!

This whole idea sounds like a little like a solution in search of a problem on some levels…



Sent from some crappy device using Tapatalk

[Timmy Crowne]

This is very interesting territory, though. I haven’t heard of any process or device that can impart the micro-dynamic signature of one signal on another. Sidechained compressors, expanders and gates are obviously the closest that come to mind. But imagine syncing multiple, calibrated instruments and reverbs to precisely follow the same curve. Not sure we’ve ever heard anything like that.

[selig]

This is very interesting territory, though. I haven’t heard of any process or device that can impart the micro-dynamic signature of one signal on another. Sidechained compressors, expanders and gates are obviously the closest that come to mind. But imagine syncing multiple, calibrated instruments and reverbs to precisely follow the same curve. Not sure we’ve ever heard anything like that.

Old sample editing apps like Alchemy could do it, but not in real time (called "Trace Envelope" and "Paste Envelope"). I can say that at least with the offline version, it was not as useful as I was hoping it would be! It's more useful if you impart a very dynamic signal onto a very non-dynamic signal, but in the end it always sounded like side-chain gating to my ear.

[RobC]

Thank you, Ladies and Gentlemen for all the great responses, discussion!

Let's see:

I totally recommend equalization/filtering before distorting the parallel signal for some delicious flavor.

I used to try the compressor's CV, but wasn't overly happy with the result. ~ Even though I created some lookahead-delay-compensation.
I remember once being desperate about mastering music so that it achieves the idiotic loudness war audio levels, and then tried to somehow add the dynamics of the good mix back. Now, the side-chain expander worked, though results were very weird. Especially since I created a multi-band expander. For a song, it sounded almost like "poor quality mp3" effect. Of course, I learned not to go to war, but make - dynamic music instead. : P (EDIT: Edited that ugly smiley out, cause text smiley rocks! xD)

Using an envelope follower might be the correct solution - I realized, that following a waveform perfectly would practically just recreate the original sound in my imagination. Yet I'm after the distorted sound's character, but with at least the flavor of the original dynamics. Not a fan of squashed/sustained effect.

A fast envelope follower could actually do the trick - especially if I do a multi-band envelope follower. Hmm, might have some EQ-like effect added, though. Could be that doing it without splitting sounds better. Only one way to find out. Not sure if wavelength matters to the envelope follower - might try converting the sound's "pitch" to 50% "speed" aka. simple slowdown... really unsure if it matters with digital waveforms (sound).

Saturation and dynamics is kind of interesting. In another discussion, I said I rather want to soft clip extreme peaks, so they don't get just squashed like with a limiter, but remain still audible, being scattered in the frequency spectrum. In this case, distortion sort of keeps the dynamic thing that was going on audible, but there's a lack of proper audio level / amplitude changes. Thus I realize, even a softer envelope following would do the trick - at least it makes sense in theory - again, one way to find out.

Now, I rather want to distort the whole sound - distorting individual bands gives a different sound.

Correction, I do want to trash the signal, so to say, xD but add dynamic changes back.

Hmm, gotta look into that vocoder - totally forgot the envelope following feature.

What's that about inverted saturation? Sounds interesting.

Yeah, summing it up, it makes sense now to rather just add a flavor of dynamics back. The distorted signal itself (even with some expansion), or if I'm lazy, the final mixed signal, will probably need some treatment anyway, since it changes the frequency spectrum quite a bit.

Now, gotta find out which device would be most suitable for expansion with an envelope follower ~ compressor, vocoder, pulverizer or scream, but not really thor then.

P.S. Let me guess, there were no Ladies involved in this discussion?