Analog modelling

[VariKusBrainZ]

To me analog means not just the bandwidth but also the non linear nature and how swapping out components can change the character of a circuit

Yeah, sure... "Digital" is always linear and swapping out components never changes the character/sound of an algorithm. Phew...

---
Just to join sarcastic opinions already expressed above. The subject is just about searching for an euphemism for words like "(all uppercase) the best sound ". Obviously one can't put "the best sound" to cover his thing (that would be too silly) so his looking for some of these crappy-hypo-buzz-words (these days every dumb 2+2 goody has the "analog sound" sticker on, so that stopped to mean anything a while ago).

If you say so

[Audiority]

Analog modelling / Virtual Analog are not buzz words. Let's take a vacuum tube as "simple" example. If you use a tanh, which will gives you a saturation/distortion similar to a valve, that's not modeling because a tube is not a wave shaper, but it's a very dynamic component. Modeling a triode requires solving some equations that will be updated over time, in order to mimic the physical behavior (and the response) of the component. There are several methods.. some effective, some less effective, but all of them take in account to replicate how currents and voltages flow in the circuit and return the result. That's modeling of just one component.

My 2 cents,
Luca

[Architeuthis]

If you use a tanh...saturation/distortion similar to a valve...that's not modeling because a tube is not a wave shaper...Modeling a triode requires solving some equations that will be updated over time, in order to mimic the physical behavior (and the response) of the component. There are several methods...all of them take in account to replicate how currents and voltages flow in the circuit and return the result...

That's exactly the kind of thinking I've been trying to express, but I don't know enough about how things work to speak in terms other than beliefs and opinions. I'd like to learn more about this kind of thing to inspire my work. I don't know any current/voltage-flow equations, but I just try formulas that update over time.

Ok, for example, I had a random thought. I keep saying digital synths should be made with physical limitations in mind. With that, I got a brilliant idea. For hard sync, instead of jumping to 0 phase instantly, put a filter, make it jump to 0 phase slowly. Boom! I got very analog-sounding soft sync. This process of idea and [useful] result happen over and over for me. Maybe this stuff is rudimentary to you, but I'm discovering these things based on pure inspiration.

I think what I really need to do is move away from trying to claim I'm creating something analog-esque and just skip the middleman. I'm trying to create acoustic / natural / physics-based sound, and after all, analog synths were invented for that purpose in the first place... Right? Or am I wrong? [people wanted synths, new-fangled-technology, to replace orchestra musicians] Filters were originally designed to solve real-world physical problems.... riiiiight? But then MS-20 came along and said filters should be screaming chaotic beasts, and we forgot that filters were meant to...uhh... filter.

Edit: After trying the idea of filtering phase, I decided to just blatantly filter the phase of the oscillators at all time for my chaos generator experimental project. Results:

[Max M.]

Analog modelling / Virtual Analog are not buzz words...

... in a perfect world. IRL, when we test some statistically valuable amount of (for example) EQs (regardless of their price/manufacturer-popularity) we can see that almost every second thing pretending to be "analog bla-bla-bla" is just an ordinal RBJ + some (usually static) saturation + a fancy analog-ish GUI. So even if *you* (+ many other developers) do care of the actual meaning of these words, the other half of manufacturers *don't* and it's all just about pure marketing hypo: "more analog -> more buzz -> more sells/downloads".

[matt42]

Analog modelling / Virtual Analog are not buzz words...

... in a perfect world. IRL, when we test some statistically valuable amount of (for example) EQs (regardless of their price/manufacturer-popularity) we can see that almost every second thing pretending to be "analog bla-bla-bla" is just an ordinal RBJ + some (usually static) saturation + a fancy analog-ish GUI. So even if *you* (+ many other developers) do care of the actual meaning of these words, the other half of manufacturers *don't* and it's all just about pure marketing hypo: "more analog -> more buzz -> more sells/downloads".

I guess the problem with a term like analog modelling is it's vagueness. How accurate or meticulous does the model have to be before the DSP oversight comity puts it's seal of approval on it. Also some people's expectations shift so the term is only considered valid if the level of accuracy is one that is practical on today's machines less so on those of 5 or so years ago, for example.

We could look at the cookbook BLT EQs and intuitively say the term analog modelling is an over reach. Perhaps the dev would respond that the whole circuit was analysed, an analog transfer function derived and then transformed and as such the term model is justified. They may have a valid point, but I'd still call bull shit. Point is the term is just vague.

[IncarnateX]

Yeah, sure... "Digital" is always linear and swapping out components never changes the character/sound of an algorithm. Phew...

Yes, the digital sound is always "sterile" myth. Strange. I have two wave lfo's and one step lfo available for each patch on my Fantom G + an "analog" function that inserts all kind of "errors" and variations into the sound but really relies on exactly the same principles that the lfos do. The lfos can be applied to each tone layer independently or all at the same time. Choosing the right values and destinations for these lfos can turn any sterile sound into something that continually varies and evolves to a small or large extent. Jeez, doesn't anyone spend time programming their synths anymore?

Btw: To me an analog sound is what comes out of an analog synth, which at minimum requires VCO/DCO and an analog filter. Period. Need no more definitions than that. And such sounds are highly varied between different analog synths. Imo, there are no universal analog sound that covers some magical common denominator of all analog synths that should make these more "organic" or "alive" than a cleverly programmed digital synth.

And it seems to me that when people want their synths to sound "analog", they want them to sound like a minimoog, a Jupiter 8, a Prophet 5, an Oberheim or some other classic, which certainly don't cover the scope diversity among analog synths. There is really no use for universals in the quest of emulating such classics. Treat them independently and you are on safe ground.

[Architeuthis]

Imo, there are no universal analog sound that covers some magical common denominator of all analog synths...

touche.

...that should make these more "organic" or "alive" than a cleverly programmed digital synth.

wait a minute, if you're just talking about the magical word "analog" then fine, but there are ways to make synths more organic and alive by design, of course now I'm using more magical words like "organic" and "alive". By organic I mean that changing one parameter may affect the overall sound beyond the one parameter you are changing. By alive I mean a synth that doesn't repeat the same waveform over time or over frequency, but maybe the timbre changes or there's a glitchiness if you do this or that. I thought THAT would allow me to use the term analog-modelled. But I guess not.

So, what is analog modelling? We've established that it has something to do with looking at components and using equations that are solved over time rather than a curve function, based on flow of electricity. Here's the words I'll be using, as this thread has helped me determine.

Analog-esque, Analog-inspired, Analog-style.

[quikquak]

How about, ' A crappy imitation of an old wooden box of wires your Dad likes...'
sorry, not helping much.

[antto]

wait a minute, if you're just talking about the magical word "analog" then fine, but there are ways to make synths more organic and alive by design, of course now I'm using more magical words like "organic" and "alive". By organic I mean that changing one parameter may affect the overall sound beyond the one parameter you are changing.

and exactly which parameter do you mean?
where would you apply this? you're not going to sprinkle this around like spices on your pizza, right?

By alive I mean a synth that doesn't repeat the same waveform over time or over frequency, but maybe the timbre changes or there's a glitchiness if you do this or that.

which synth does that and in what way exactly? why is it important to be modelled?
"maybe" the timbre changes? well, does it change or not? are you going to use the std::maybe object?

I thought THAT would allow me to use the term analog-modelled. But I guess not.

nothing stops you from using the term
but if you choose a specific synth to model, then you have the ability to perform A/B tests
while if you just put some algorithms together while wearing an "i love analog" t-shirt, you have nothing to compare against.. unless you make an analog circuit based on your digital model (which is possible, but can you? )

[Architeuthis]

are you going to use the std::maybe object?

Uhh, hell yeah, I'm always using std::maybe. For one example of "alive", defined as "unpredictable timbral changes and/or glitchyness over frequency". This is pitch-tracking FM (as most FM should be), except using a fixed frequency lowpass filter on the modulation which then creates vibrato on some notes, growl on others, and for the lowest notes we get a strong clear FM PLUS vibrato. Edit: Oh also, there's pitch glide, which affects the FM which causes the glide to unpredictably not land fast enough on the target note:

.

For an example of "organic" defined as "changing one parameter affects the sound beyond the one parameter", an ADSR with a "pluck" slider would change the ADR times (yes attack time is important for pluckyness!) based on an algorithm. That's very intentional design. How about a waveform that becomes noisy as it is pushed into high frequency? Also, any chaos/feedback patches will change drastically for one parameter change. Also, any time you just blatantly couple multiple parameters as a design decision or as an emergent behavior... I'll have better examples of this in the future, but here's my attempts at designing an analog-inspired filter. It becomes less stable at lower frequency [or at high frequency], so frequency and stability are coupled by emergent behavior:

[BertKoor]

All really nice features for a synth, but I would not expect to find any of those in an analog synth.

[antto]

so you basically completely ignored the important points in my post, and just quoted that std::maybe joke and used it to throw in some videos showing.. something..

i suggest you re-read the previous posts... there might be some answers to your very questions which you overlooked

[Architeuthis]

i suggest you re-read the previous posts... there might be some answers to your very questions which you overlooked

Please be more specific. I just re-read your post. I don't know what question you are referring to, also I feel that I've fully answered your queries, I don't know what I missed.

All really nice features for a synth, but I would not expect to find any of those in an analog synth.

Most of what I do is possible on an analog modular. But, I'm starting to lean toward using "analog-inspired" to describe my work, it's not any specific synth that inspires me, it's the underlying nature of how analog synths use a tangible thing called electricity flowing through components that can't react perfectly linearly, perfectly instantly, etc, it's difficult to be more specific than that. Edit: Also I take inspiration from acoustic instruments / physical world.

[antto]

i suggest you re-read the previous posts... there might be some answers to your very questions which you overlooked

Please be more specific. I just re-read your post. I don't know what question you are referring to, also I feel that I've fully answered your queries, I don't know what I missed.

then i must have given answers to your questions which have been rhetorical
i'll just give you a suspicious green flag to go on and call your ideas "analog" and/or "natural", as long as that makes you feel good
users will not be mislead by marketing terms for the first time
they are used to it

so.. it's gonna be like "this is an accurate model of a hypothetical analog modular synthesizer patch, with very simple dsp algorithms and very little CPU usage"
sounds exactly the same as good old "VA"

[antto]

But, I'm starting to lean toward using "analog-inspired" to describe my work, it's not any specific synth that inspires me, it's the underlying nature of how analog synths use a tangible thing called electricity flowing through components that can't react perfectly linearly, perfectly instantly, etc, it's difficult to be more specific than that.

a toy-like monosynth from the 80s has a quite decent and usable VCO tuning across its 5-octave pitch range
one of your videos above demonstrated something that was either "WTF" or the faint presence of a familiar melody burried in a forest of out-of-tune-ness

as the more wise said earlier: there are good sounding and bad sounding analog synths..
people often think that the reason for the good ones to sound so good is that they're analog, while forgetting that the bad ones are also analog
you know.. electricity flows thru the Volca Bass too, but it just doesn't ring my bell.. at all
and i've heard a bunch of patches created on big modulars which also did not impress me at all

the wiser folks already went mute here, i'm still replying to you because a bunch of years ago i was pretty much in your shoes..
i was working on a bunch of very strongly analog-inspired synths (aka not based on any-even-remotely-thing), and talking about it on forums, sharing audio demos during the progress..
users were very enthusiastic, the more experienced ones were sceptical, and i thought they were just attacking me out of envy or something
at the end, those projects went nowhere