Thanks for sharing, and I took a quick look, but don’t quite understand it.
Coincidentally, I recently came across, possibly on KVR, an old comment of Urs’ that I interpreted as a possibility at least that some, much, or all of Zebra is really just a glorified granular so-called synth.
Perhaps this is not the case, but if it is, it might go some ways to explaining, for example, why Zebra and other synths along its lines might be especially easy on the CPU.
I have also read that granular synths aren’t really synths, just manipulators of tiny audio fragments-- in short, ‘samples/samplers’.
There’s what Vital calls “wave source” that’s actually additive harmonics. Then there’s audio file source, which is sliced and diced sample. Then there’s “line source” that’s like the first one but without the control of individual harmonics.
Ig line and wave source both get rendered to a sampled wavetable before put out, otherwise Vital would likely eat tons of more cpu time to the point where my rig could barely run it, at least if it would be anything near Zebralette3 which can choke mu cpu with one chord with its additive engine. Dunno how Vital handles the spectral distortion effects though, ig to make sure you gotta read the source code.
Resynthesis makes an audio file and uses that as an audio file source.
That’s like the definition of a granular synth. It doesn’t make them any less of a synthesizer. The source doesn’t matter, what a synth does is that takes audio source, manipulates it, and puts it out according to user defined parameters. Synths are basically effect chains with primary sound source.
What sets them apart from samplers imo is that a sampler plays back samples that can be used as they are, where a synth uses sampled material to construct a new kind of a signal. The line seems blurry to me, since surely one can use many samplers like a synth, but rarely can use a synth like a sampler.
AFAIU, granular ‘synths’ don’t reinterpret the audio, though. They use it as-is and just cut it up/spit it out in fancy ways. Even AI seems to be properly reinterpreting the audio.
As per your previous comment, it does seem like that, like Vital is kind of what I might feel is cheating/kludging/‘hybridizing’ its way to sound design and isn’t really approaching it from a kind of ‘purist’ standpoint, say by using 516 oscillators to produce 516 sine waves and that’s that.
Maybe I should be looking at Virsyn’s upgraded 64-bit Cube 2 or Razor or Parsec?
What do you think of physical modeling synths? Seems some approach ‘physical modeling’ differently too.
If we want to talk effect chains, seems granular synths are glorified equalizers, yes? But again, my feeling about it, at least so far, is of granular synths as not really being synths in a pure sense, but more like fancy samplers because they are using the same audio.
The differences can be subtle but probably best explained by an impersonator impersonating your voice, say. It’s not your voice anymore, but a reinterpretation– a synthesis – of it. To me, that seems like real synthesis.
There’s what Vital calls “wave source” that’s actually additive harmonics.
When and where is that applied and could one work entirely in that part/section/realm in Vital without touching the rest? If so, how flexible is it?
It would be just impractical to play back 512 or 1024 sines real time to construct a waveform in terms of processing economy. If your use case requires that ig you gotta use that sparingly, bounce it or wait until we have more powerful processors to be able to use that kind of synthesis extensively in production. Playing a single synth live might work okay if your rig has enough processing power, but we’re really hitting the limits here.
Other than that there might be synths that use trigonometric maths to generate certain waveforms, but they’re then bound to the limits of those.
But I just mentioned 3 additive synths that have been around for over a decade-- Razor, Parsec and Cube 2. I’ll add Harmor.
So why should current rigs not be able to handle what rigs could ~10 years ago, and therefore, by implication, Vital, if I can use it strictly in the additive sense? Can I? Did you not just write, “There’s what Vital calls ‘wave source’ that’s actually additive harmonics.” ?
Wave source is one of the wavetable editor’s primary sound sources. As said, that likely does get rendered to sampled wavetable frames non-realtime as reaching the nyquist frequency from the bottom of human hearing would require about a thousand additive sines that would eat up so much cpu time that it’s hard to justify that with the current hw tech. As said, you can get a taste of that with Zebralette3.
Still a bit funny that you’re not in the analog camp when discussing purity and holism of sound reproduction - not even a mention of euro racks??
I joke ofc, @glomerol id recommend getting some books on the mathematics and design behind sound reproduction, it’ll help answer some of these questions
I’ve done some deep dives myself but in the end I decided that time spent researching is less time creating music
Whats more important, the tools used to create or the creation itself? An age old art conundrum, but that’s the best part - it’s art - you get to make up questions and make up your own answers. My tip is to pick and synth and start making music (or sound design, patches, etc whatever your thing is)
Don’t let holism consume you Aristotle, unless that torment searching for your answer makes some sick music
since all digital audio is a sequence of sample values, the entire question is kind of moot in that the issue lacks a genuine case of controversy. even if vital generated only pure math sine waves, it would be doing that in an inherently artificial manner, by supplying samples one at a time instead of groups of samples such as wavetables. both ways, they are still samples.
similarly, if you only had analog synthesizers to use, it would take a lot of allegorically boolean operations for electrical engineers to synchronize the oscillations to arbitrary time constants. the binary component would be in the form of design decisions that were based on discrete measurements probably in the decimal system using standard references. and if you look at what any engineer is doing in an abstract way, they are programming, or conducting a series of events or forces that attempt to arrive at a goal in a repeatable and predictable way. otherwise that synth would really be garbage if it were up to a vote.
also, as far as whether a synth is additive with a zillion sine waves or not is also not a real argument because at the end of the day, the only math a computer ever does is addition. every math function is a series of addition operations. so if you want a program to go fast, you need to look at those operations and figure out a way to get the same or a similar result at the end with as few operations as possible, because at the end of the day you can go crazy and make a “perfect” filter that contains all the quirky non-linearities in the known universe, you’ve just made a plugin that is a gigabyte in size and the latency of 60 seconds on a threadripper, and it wouldn’t sound any different than a synthedit filter from 20 years ago to most people’s ears, so there’s such a thing as diminishing returns.
heck, the same kind of arguments are going on right now probably at gearslutz about preamps and microphones. so, why not just try to focus on writing your sonic story and not worry as much about whether the tool is in the way. tools are like typewriters. what story are you going to write on your typewriter?
Perhaps, at least for some/me, it boils down, at least in large part, to how and how well the synth can echo reality and/or its complexities and/or somehow transcend them. ‘The math’.
See my most recent main topic post, which is a bit of a continuation/elaboration of this one, about the Terrain synth and let me know what you think. Better still, before you do, try it. I haven’t yet.
Oh absolutely, I don’t deny that. But I mean, I’ve listened to many presets from various wavetable synths-- almost all of them actually, including Vital’s, and while they are all fine and nice, they seem to somehow be lacking in, again, vitality. As much as it might annoy you to read. LOL <3
…Also, I’ll add-- and concede that this will also risk annoying you-- that some have suggested less of a vitality in Vital’s development cycle and that Vital seems to have this weird sort of half-baked FLOSS/NON-FLOSS aspect to it to boot. On the other hand, Terrain is FLOSS, and possibly under the GPL as well. Ditto with Vast Dynamics’ Vaporizer wavetable synth, now that I think about it.
The market’s saturated with subtractive and wavetable synths, maybe oversaturated. So I like how Terrain has sort of broken out of that box a little. I think there’s a lot its dev (and others because it’s properly FLOSS) can add to it.
My friend, your question is hard to answer. But I’ll try to give my insight as an artist point of view (Most of you share common artistry): It was morning and I saw the sun coming up. I imagined a 4 synth ensemble, with no percussion, but sounds - a distorted violin, a not-gated crunchy bass (like those you listen when Skrillex), and as the sun comes up the music was louder and brighter. I opened Vital back home, and still gathering the most accurate form to bring this imaginative sound to reality. It’s still in my head. But, do you think I would ask an A.I to transcribe what I “listen” ? No. Catch similar samples at FreeSound? No. I hope I don’t miss the point here, but, in my opinion, Vital could be compared to a drawing tool where you have the best pencils and brushes. Nevertheless, sample exists in its 4th head - and it’s limited. You have to edit the sample first in Audacity, ‘cause once in Vital the sample editing is crippled. Moral story: It’s a true synth. Perfect for SounDesign. An odd, presets are too harsh, but I understand that it’s Vital essence. Tip: Try to get favorite presets and reverse engineer it. Sincerely, Marcio Hendrik (@artigoaudio)
when analog synthesizers first came out, they were trying to artificially reproduce violin, pipes, keys, etc, that’s why they are called synthesizers. and really they did a crappy job at trying to synthetically reproduce them. they somewhat accidentally ended up being their own esthetic and ironically people started trying to synthesize synthesizers but then they ended up creating a new esthetic again. it’s as if nostalgia is inspiring people to reach back and remake things, they kind of fail, and accidentally create something new. If it floats your boat to want to reach back and reproduce something that’s fine, but remember a lot of great new inventions happened by accident while trying to do something else.
a good example of this is the jake paul mike tyson fight. a lot of us wanted mike to make mincemeat of jake’s face, but what ended up happening was a kind of cinematic hybrid of night of the living dead and brokeback mountain, and i came away from it feeling ashamed to have wanted them to hurt each other and how boxing turns the audience into sadists and maybe the sport of boxing should be banned. all because of their attempt to synthesize a boxing match.
i mean a wavetable is a bunch of waveforms stored, where each frame is a ‘snapshot’ of the signal that will be sent out to the speaker. thats basically it. in a digital system, it doesnt really matter whether you do that via fm/additive/subtractive or by wavetable, or anything else. its all going to be effectively producing the same outcome, because of digital being digital, which uses logic gates and cpu cycles to create speaker output.
so vital is no different than any other digital synth, aside from its features ofc, it is technically producing the exact same samples as any other digital synth.
Wavetable synths fall into synthesis because you completely alter the sound into a creation whereas a rompler is like a preset that you can change just a little.
A rompler would be like the E-mu Proteus or Roland JV series. Nexus 1 & 2 were romplers even though they tried to tell people they weren’t, it may have had synthesis capabilities under the hood but those were not accessible by users.
A wavetable could be considered similar to a sample, but a sampler is not a rompler either because you can manipulate the sound to a great extent., A wavetable it’s closer to your standard waveform like square, sine, triangle, and pulse. A wavetable on it’s own usually doesn’t sound great just like a traditional waveform until you start adding filters and sculpting the sound.
An idea of an additive synthesizer seems easy enough: take sine waves, add them, get the output signal. But when you go into digital sound processing, you learn that computing lots of sine values for each audio sample is just unnecessarily costly, even on current hardware. So instead one either uses recurrent formulas that are in effect just resonant IIR filters, or, yes, uses resampling if the sum of all the sines is periodic—first generating a waveform for a suitably low pitch and then using it with very cheaply to computation, if compared to a giant stack of filters each for its own frequency.
Alas, the fact is, resampling can reproduce waveforms extremely well if implemented using many long-known improvements over dumb nearest-neighbor method when you just pick whatever value the sample has at the phase nearest to the phase you’re interested in. As resampling artifacts go, both harmonic and inharmonic aliasing are well-known how to deal with (starting with oversampled generation, highpass filtering and then downsampling—but there is more). There are other resampling artifacts and I don’t remember about them right now but they are also not a mystery regarding what to do about them.
So in the end it’s not hard to have good resynthesizing oscillator which is equivalent to blind summing of mathematically pure sines sampled at the output sample rate. Or at least equivalent up to noise well below any sensible level (say, −200 dBFS). That’s why I would think the OP question as a kind of suspect audiophile stuff. Sorry.
