U-HE Zebra 3 Alpha Prototype Developments

[HcDoom]

Do I read this wrong or there are people around here who really think they can teach Urs how synthesis works?

[liquidsound]

^^^

[Daru925]

Do I read this wrong or there are people around here who really think they can teach Urs how synthesis works?

All your waves are belong to urs.

[kmonkey]

Do I read this wrong or there are people around here who really think they can teach Urs how synthesis works?

You are on KVR. Here, people consider themselves "experts" in every field right after seeing one Youtube video from yet another buffon with yet another "game-changing" hyperbole thrown out.

[Urs]

Do I read this wrong or there are people around here who really think they can teach Urs how synthesis works?

LOL, I have been wrong at times

In any case, Zebra's oscillators work in ways that are not easily explained and that do not have many siblings out there. I think it's natural for people to try to compare them to technologies that are more commonly known, such as wavetable synthesis or resynthesis. However, the pros and cons of these technologies do not necessarily map, i.e. they are not necessarily comparable. As a result, people sometimes request the pros of other technologies for Zebra even if those pros can not be implemented. They also sometimes find the cons of other technologies in Zebra and tell us about synths that offer alternative solutions, even if those synths use different synthesis methods.

I think that's quite natural. We offer a pretty straight forward user model in Zebra (draw lines, apply effects, waveform changes accordingly) to something that is very, very abstract and much more complex under the hood. We can't expect people to understand the abstract part as the abstract part is something we can't represent visually, nor conceptually as long as we want to retain our little synthesis secrets.

[zerocrossing]

Do I read this wrong or there are people around here who really think they can teach Urs how synthesis works?

LOL, I have been wrong at times

In any case, Zebra's oscillators work in ways that are not easily explained and that do not have many siblings out there. I think it's natural for people to try to compare them to technologies that are more commonly known, such as wavetable synthesis or resynthesis. However, the pros and cons of these technologies do not necessarily map, i.e. they are not necessarily comparable. As a result, people sometimes request the pros of other technologies for Zebra even if those pros can not be implemented. They also sometimes find the cons of other technologies in Zebra and tell us about synths that offer alternative solutions, even if those synths use different synthesis methods.

I think that's quite natural. We offer a pretty straight forward user model in Zebra (draw lines, apply effects, waveform changes accordingly) to something that is very, very abstract and much more complex under the hood. We can't expect people to understand the abstract part as the abstract part is something we can't represent visually, nor conceptually as long as we want to retain our little synthesis secrets.

I remember when I first started learning FM, going from sine wave to a sawtooth and back, and thinking, "oh, like a traditional analog with an oscillator and a filter." But of course, not. There are many ways to reach a similar result, but also a lot of area where the circles in the Venn diagram don't overlap. The fun for me happens when I exploit those places. As an end user, I try and not think about the underlying technology that much, because I don't find it particularly useful. Learning how the control inputs effect the results is where the rubber hits the road. Everything that's happening inside the computer is math, which I am notoriously bad at. I am generally pretty good at learning a synth on a high level, and that's enough to give me the results I'm after.

[seafire]

. Everything that's happening inside the computer is math, which I am notoriously bad at.

It's very basic maths. One and zero, yes/no, on/off.

Primary school stuff really.

[kmonkey]

It's very basic maths. One and zero, yes/no, on/off.

Primary school stuff really.

That was a really stupid thing to say. Oversimplification. No different than me saying that the creation of human life is very easy to explain. Primary school stuff. You just have to have male and female, and they need to engage in notallowedtosay relationship and we have offspring.

Yet, to this date, even the brightest scientist can not explain the creation of life and can not repeat "primary school stuff" in labs. I mean, they can when they are not working hihi:

[e-theory]

It does not have wavetables in the common sense.

(it has splines based curves which use a single cycle wavetable format as an intermediate storage format for further processing in the spectral or time domain)

Will you be opening this format for people to be able to create their own tables for Zebra 3 Urs?

[e-theory]

Do I read this wrong or there are people around here who really think they can teach Urs how synthesis works?

Yes, you are reading it wrong. gentleclockdivider said I was "doing it wrong" then disappeared when I posted the patch where he could see it's not actually possible to do what he was saying.

As for the rest of what I posted, Urs was just giving me the technical details, which I really appreciate. I certainly wasn't "trying to teach Urs how synthesis works". That's a weird take. And the reason, as he confirmed, that Zebra-2 doesn't smoothly interpolate the spline points when modulated, is that it's basically just cross-fading wavetables like most other wavetables synths, not moving the points of the splines (except when it generates the wavetable entries). That's totally fine, and totally explains what I am seeing. The entire reason for posting was to ask if Zebra-3 will be implemented in the same way, and it sounds like it'll be slightly different.

I am a mathematician and computer programmer by trade, so I appreciate Urs explaining the technical details.

[Urs]

It does not have wavetables in the common sense.

(it has splines based curves which use a single cycle wavetable format as an intermediate storage format for further processing in the spectral or time domain)

Will you be opening this format for people to be able to create their own tables for Zebra 3 Urs?

It does not have tables. It has curves that seamlessly morph between shapes - not just for oscillators waveforms and spectra but also for envelopes, and phase modulators, freely shapable filters etc, within the oscillator effects.

People can create their own waveforms in Zebra 3. The whole point (almost) about Zebra 3 is that it has editors which lend a lot of functionality from CAD and vector graphics applications. Hence, people can create their own things, but they are not wavetables or tables of any kind. They are keyframes that define an otherwise "liquid" set of states in-between.

People can export morphing oscillator waveforms to wavetables compatible with Hive and Serum, but the internal storage format, processing etc. has little in common with wavetables used by Hive or Serum. What the export function is, it samples the morphed states a hundred times and saves the waveform converted to a wavetable.

[e-theory]

Thanks Urs, the reason I ask is that there are cases where generating the files algorithmically might be more desirable than using the interface you provide. Say when wanting to experiment with mathematical curves, or progressions of control points on curves that might be difficult or time-consuming to draw manually. In these cases an open file-format could be useful.

Unless of course Zebra 3 provides an expression-evaluator like Dune 3 does, where you can type in an expression and it will use this to generate the waveforms.

[Urs]

Hive has an expression evaluator (well, a language to create and manipulate wavetables). This is called ".uhm" and it's documented on Hive's website and user guides.

The spline curves in Zebra 3 can be copied and pasted as .uhm text bits. At some point we're going to add evaluation of these to Hive's .uhm parser, but this is currently only available internally for us. With it, one can create Zebra 3 curves from wavetables and vice versa.

Zebra 3 can also copy and paste curves back and forth in .svg format, as a bezier path. This is open source and if you wish to use mathematical expressions to create bezier paths, you can totally do that today - and use them in Zebralette 3.

[Urs]

(I have a hunch though that we're still discussing Zebra 3 as if it was a wavetable synthesiser where somehow arbitrary data can be used as a waveform. But this is not the case. The strength of Zebra 3 lies within the abstraction of spline based curves, which have the advantage of seamless morphability. It will always be better to use few points and high controllability over complex waveforms derived from sources that are less suited for morphing and control through human interaction.)

[Urs]

That said, people have written waveform importers for Zebra 2, but afaik they were always based on wavetables.

As .uhm is text based and thus Zebra 3 curves/waveforms/envelopes can be copied back and forth between Zebra 3 and a text editor, I'm sure people can figure out easily what those curves are made of. It's rather plain cubic bezier curves, really, there is not much of a need to document or open source anything. Naturally, unlike the ones used in Photoshop or Illustrator, those curves may only ever have one Y value for each X. That's the only limitation.