Saturation Types

I'm deferring to the talented minds that frequent this place as this is a subject beyond my ken.

I love digital audio the quality is far in excess than the best top end tape machines that I've had the pleasure of having access too.

But tape, valves (tubes) and quirks in circuitry did certainly tinge the clear water in a nice way most people would agree, to the extent there is a massive market in colouration plugs trying to emulate this phenomenon.

Of yore I've rarely found anything that does the job and gives the 'feel' that recording to tape used to give me, recently though I've found a few things that have come close to satisfying that need to colour a sound, not necessarily to emulate a hardware classic, but to add a certain richness, especially at the point where tape would start to over-saturate.

I'm not interested in the curses of wow and flutter and hiss that used to plague analogue gear, everyone used to hate it back then so I don't see why its elimination should be considered a bad thing I love having no hissing, wobble and bleed through. I don't want to emulate a cassette player, nor do I have a need to 100% emulate a boutique classic, I just want the good stuff working in a digital realm that works.  I'm currently loving the freeware Ferric TDS plug which while not following any classic hardware allows adjustment of compression and saturation algorithms which can then be taken toward the brink of where tape would start to distort through over saturation.  And saturation to me doesn't mean dirty distortion it just means being able to take it to that threshold and stop short of it.

Here's a guy I've been enjoying the work of and use a few of his things and like the results, and I'm not after a the plug-in I use is better than yours type thread, else I'd have posted this in the X1 forum.  I'm more into the considered views of those that come here on the current state of saturation in the digital realm.

To kick off a discussion here's Bootsy's views on stateful saturation instead of just using a look-up table of results from a vintage box, as I think this is the current thinking that is causing an upturn in good solutions becoming more available.

http://varietyofsound.wor...s-stateful-saturation/

I found it interesting that someone who is thinking carefully about this spoke about the nature of saturation and didn't resort to using words like "tubes".

Yes there's plenty more good commentary he makes on the subject besides, as well as some good links out.

The key area for me is being there is an absence of a digital ceiling before flat-top clipping occurs you start getting a nice warning with his stuff particularly TDS that the point of over-saturation is about to occur and you can take your signal into that zone with variable amount of saturation and or compression already dialled in.

For me digital is always going to benefit either from the use of external boxes or this kind of thinking that isn't going to just going to mimic a few attributes of one of those boxes in pre-defined circumstances.

We want active behaviour which is how I understand stateful, like an amp in it's warmed up state, or dealing with a high frequency transient when the state of an amp say has just had a drum kick pass through it and when it is in a different state dealing with a similar transient  in a quiet passage.

So yes in order to be alive the algorithms must have some awareness of the qualities of the signal flowing through them and react accordingly rather than just pulling plotted points from a pre-defined look up table that may or may not follow some vintage gears output results.

Here's a guideline about judging saturation effects Bootsy posted awhile back, for anyone interested in playing around with this stuff there's some free downloadable tools too.

http://varietyofsound.wor...ng-saturation-effects/

I'll try and get some before and after waves of some affected signals too.

My take on stateless versus stateful in this context: in software, the terms mean that the code either keeps track of previous data values or how the data got to where it is now (stateful) or only looks at where the data is at any point in time (stateless).

HTML-based applications are generally considered to be stateless, because each page you display is an autonomous entity that knows little or nothing about what you did prior to displaying that page. From a programmer's point of view, it's a challenging environment if you want to introduce any kind of intelligence in the code. Cookies are an attempt to provide some statefulness, but they're very limited. None of this has anything to do with saturation emulation directly, it merely illustrates the concept of "state" in software.

I think what Bootsy is referring to is this: magnetic devices aren't just simple compressors, but are frequency-sensitive, especially transformers. A transformer's reaction to a given signal level is 5% about its absolute value and 95% about how quickly or slowly it reached that level.

Once you introduce frequency into the equation, you are necessarily talking about comparing values over time, deciding what to do to a particular sample based on what's gone before. You can't do that by simply looking up a transform in a table, which apparently is how most hardware emulations do it.

It would seem that Bootsy's approach to making saturation emulation stateful is an attempt to refine the emulation to more accurately represent how a real magnetic device acts. Users have long complained that digital saturators don't sound quite right, which suggests that the model needs refinement.

Speaking of refining models...one of my customers back in the day was the National Weather Service in Anchorage, Alaska. One of the meteorologists there told me that it is theoretically possible to be 100% accurate in weather forecasting, but the reason we can't ever be is that a model sufficiently detailed to achieve complete accuracy would take too long to crunch the numbers. The weather would have come and gone before it could be predicted. This is the same problem facing plugin developers, only in a much shorter timeframe.

bitflipper


My take on stateless versus stateful in this context: in software, the terms mean that the code either keeps track of previous data values or how the data got to where it is now (stateful) or only looks at where the data is at any point in time (stateless).

HTML-based applications are generally considered to be stateless, because each page you display is an autonomous entity that knows little or nothing about what you did prior to displaying that page. From a programmer's point of view, it's a challenging environment if you want to introduce any kind of intelligence in the code. Cookies are an attempt to provide some statefulness, but they're very limited. None of this has anything to do with saturation emulation directly, it merely illustrates the concept of "state" in software.

I think what Bootsy is referring to is this: magnetic devices aren't just simple compressors, but are frequency-sensitive, especially transformers. A transformer's reaction to a given signal level is 5% about its absolute value and 95% about how quickly or slowly it reached that level.

Once you introduce frequency into the equation, you are necessarily talking about comparing values over time, deciding what to do to a particular sample based on what's gone before. You can't do that by simply looking up a transform in a table, which apparently is how most hardware emulations do it.

It would seem that Bootsy's approach to making saturation emulation stateful is an attempt to refine the emulation to more accurately represent how a real magnetic device acts. Users have long complained that digital saturators don't sound quite right, which suggests that the model needs refinement.

Speaking of refining models...one of my customers back in the day was the National Weather Service in Anchorage, Alaska. One of the meteorologists there told me that it is theoretically possible to be 100% accurate in weather forecasting, but the reason we can't ever be is that a model sufficiently detailed to achieve complete accuracy would take too long to crunch the numbers. The weather would have come and gone before it could be predicted. This is the same problem facing plugin developers, only in a much shorter timeframe.

I think all in all you and I just said pretty much the same thing.

I differ here though, to get round the emboldened issue there have been in use from the beginning, approximations, bit depth after all is just a stepped approximation albeit one we are able to be accurate enough with in order for it not to be perceptible for all practical purposes.  Approximations therefore I don't think are the hold up here as many emulations have been shown to work well within the constraints of the small time frames that are being worked on, and approximations will always have their uses when they don't impact on perceptible quality regardless of  the nature of the algorithm being used.

So stateful to my mind is really the key watch word for future development where a signal is acted on differently dependent on its content and context, rather than the resolution of a pre-defined pattern or even multiple patterns of behaviour being imposed however complex.  This is the key as far as I can see to adding something of the living and breathing non-linearity inherent in analogue circuitry.

The main drawbacks to my mind come from trying to emulate analogue behaviour in the digital realm, for example algorithmic reverbs are notoriously difficult to implement digitally, yet conversely convolution 'verbs are ridiculously easy to implement aside from the brute force performance considerations needed to use a high resolution wav file as the predefined look-up table.

I personally think the dawn of digital solutions got delayed because of the mistake of attempting to re-create old analogue methods in the digital realm rather than using new algorithms more suited to digital which may end up having far more flexibility and producing better living and breathing results by embracing the difference between the two realms. and actually that seems to now be what is starting to happen.

I was surprised to see for example recently how many people that have never been used to seeing an analogue mixer don't even bother with their DAW's console view and even those that were used to them at one time, make out really well without it these days. Console view was just a familiar frame of reference that made it easier for those of us from the old days to get to grips with, it had nothing necessarily to do with good DAW interface design which doesn't need these days to cling to that old model.

The same thing, with some hope I think, I'm starting to see is happening with audio processing and the signs of it are starting to show up in the same time frame. 

Analogue kind of imposed that you had defined stages of processing which showed up on a console as basically pre-amp, eq, compression and limiting which then was committed to tape with various types of saturation and colouration being introduced along the chain, getting away from that dated ball and chain methodology that no longer need apply to DSP has been the hardest part of the journey from analogue IMO and I therefore expect to see some exciting new stuff in your usually excellent report from NAMM this year...

In particular I'm looking for the killer plug that handles dynamics in the frequency domain whilst applying an addictive kind of saturation whereby I can achieve world domination by the hypnosis and complete surrender to me of anybody listening to my productions so I can get them to do my bidding.

Preferably for under $25 would be nice too.

Hi, Jon,
 
A belated Happy New Year to you, my friend!
 
Hopefully on topic, just a suggestion for a free demo to try, unfortunately a bit more than $25 ($79-US), but I think it's well worth the price.  Voxengo Analogflux Suite, actually a suite of 5 great plugins, free demo available at http://www.voxengo.com/product/analogflux/
 
I use the included TapeBus plugin on almost every bus in my projects, and I'm quite happy with the very subtle difference in sound the plugin provides.
 
A blurb from the plugin description:
 
"The TapeBus VST plugin recreates characteristic elements of the reel-to-reel tape sound.  This includes saturation, modulation noise and smearing effects which are known for the 'analog' feel they bring to any audio recording.  This plug-in also applies a selected impulse response taken by us from the existing tape machine."
 
I think it's well worth a free test drive just to see what it does to one of your mixes.
 
Have a good one, buddy!
 
Best wishes,
Bob

Don't know if my dribble helps any:

I've had the same complex ordeal, Jon, at this Motel (with lofi wifi).  I'll try to listen more than speak here (perhaps later I'll actually be able to follow your URL) ... and exude a few ponderings.

Tube vs. tape vs. warm vs. digital vs. even harmonics vs. odd harmonics vs. consonant distortion vs. some new coloration ... etc.

The psycho-acoustic scientitists, fortunately, have manufactured lush 'stereotypical presets' for this feeble song-writer (moi) to hopefully rule the world with. 

Like a Vincent oil painting ... things can get a bit rich, lush, and/or 'bleeding' with color ... some eyes/ears may fancy a restful passage after such hypnotic colorations, IIRC. ... so as not to overwhelm the senses or distort things ...

IOWs, I'd better let you 'others' clear the clouds on this most noble thread.

bitflipper

I'm looking for the killer plug that handles dynamics in the frequency domain whilst applying an addictive kind of saturation whereby I can achieve world domination by the hypnosis and complete surrender to me of anybody listening to my productions so I can get them to do my bidding.

Post that, word-for-word, on KVR and see what kinds of responses you get! Someone will surely point you to their SynthEdit creation that does exactly that.

I don't doubt it...

I'm scared to do it though as somebody may have actually done it already then I'd be consigned to do their bidding...