Zebra 3 Public Beta 3 Revision 20977

[Lbdunequest]

I wonder if there will be wavetables already made in the final release. Tbh im not of those who create wavetables

[Urs]

I wonder if there will be wavetables already made in the final release. Tbh im not of those who create wavetables

There'll be Oscillator presets. Not hundreds, but quite a few.

Our heads are pretty busy with a gazillion of other things...

Zebra 3 won't start naked, but if we have anything, it's a good track record on improving our software over time.

[geronimo]

For Wavetables, I think Hive 2 is more relevant.

[benmason]

Is the next beta being released this week?

[Urs]

Is the next beta being released this week?

The team is working extra shifts to make that happen, but we are at least two days behind schedule. For instance, we need to make another few changes on the UI, which might take all of today and tomorrow, then update all screenshots in the 200 pages user guide. It's tight, and Friday is a public holiday here.

[loctune]

Ok, here's my super long review and thoughts after about a week beta testing this Zebra 3 revision. Zebra 3 is deep and I may have still been scratching the surface, but based on my personal experience, I think I've got a basic grasp of its functionalities and limitations. There are a number of things I really adore, but also a few things I feel lacking/lagging behind for a flagship synth released in 2026.

There will be mentioning of and comparison to a few other soft synths I own and use frequently in this post. I have no intention saying Zebra 3 as a product is inferior or superior to any of them. They are mentioned only so that I can talk about some comparisons and expectations. And while my post is not going to be all about singing praises, I have no intention to offend anyone nor to talk down the effort spent by the developer team in any way. Actually, I am grateful that I can beta test Zebra 3 and I would like to express my thanks to Urs for his help and his openness to communication. If anything I say here is wrong or inaccurate, please feel free to point them out.

I am happy to find Zebra 3 is not another Serum wannabe in 2026. Instead of another 2 osc wavetable & sample soft synth, Zebra 3 is something with its own design, thought and workflow. And I am glad to see Urs' attitude to define Zebra as a "pure synth" and focus on synthesis rather than get into samples. However, as Zebra's price is clearly flagship level, I also have quite high expectation about it: my common sense tells me that a flagship released in 2026 which refuses to do samples must have highly advanced pure synthesis functionalities. Turns out many of my expections are met, albeit not all of them.

Let's start with the main OSC. Zebra 3 took a 16 frame vector based approach instead of the standard 256 frame wavetable. I haven't done enough research or comparison to say which approach is better, but from a pure technical POV I can see more flexibility in the 256 frame wavetables (16x higher number of different waveshapes available in a wavetable and possibly better compatibility when importing larger wav files) while I think Zebra might have the edge on smoother morphing between different frames.

I have zero complaints about Zebra 3's OSC sound. It's definitely on par with the best software synths in 2026 if not better. The additive mode is cool and the two OSC FX allow for quite some extra waveshaping functionalities (I'd like to have a feature request for an OSC FX which creates a sub-osc based on the current one, effectively a doubler with variable pitch). However, there are also a number of things I feel "incomplete", mainly the wavetable editor: there is no curve preset (coming next beta?), and there is no easy way to create a bunch of related frames in one pass. Users have to manually draw every frame in, which is inconvenient. The lack of spectral editing mode in Zebra 3's wavetable editor is also a huge concern. Given that two points with 0 distance degrades to a single one in the wavetable editor, there's no way to draw single vertical lines in the wavetable editor representing clean harmonics, and the produced waveform is always highly distorted in spectral mode. This lacking of a useable spectral editor (the mapper editor seems like a perfect fit here) makes its spectral rendering mode almost useless, at least in the current revision. Edit: this part is wrong. You can do partials in the wavetable editor to do clean harmonics. It's only there's some inconsistent interpolations going on in the maths mode, that you may have to do only a portion of a partial grid to get the correct behavior.

Then the FMO. During the beta testing, I keep feeling that the design of having individual FM oscillators is an afterthought remedy, trying to work around the issue of non-existing OSC FM (something still under evaluation according to Urs, really hope to see it soon) in Zebra. While it works and sounds good, it's still a heavily limited solution. Both the carrier and the built in modulator only do pure sine (distorted saw with feedback), which, since Zebra 3 isn't aiming to be a DX7 emulation, makes me feel kinda unsatisfied. Even the FM8 released 20 years ago does more waveforms than pure sine and has basic phase control, so for FM-specific osc in a flagship synth in 2026, I'd expect to see at least some more classical waveforms - or even better, editable single-frame wavetable - added as carrier and modulator options.

Now something super positive. The modal synthesis is what interests me most in Zebra 3. And it does such a nice job that I have literally nothing to complain about here. The resonators are really nicely crafted. And I am happy that everything is plain CSV. Best modal synthesis I've seen in software synth.

There are a great number of analog-style filter options to choose from. And they all sound great. Yet there are only 4 single filters (some OSC FX can produce filter-like effect, but they aren't really filters) one can use in a single Zebra instance. Given that all the filter options are single filters, once you start trying something structrally more complex - let's say I am doing some MS-20 style HPF to LPF in serial or parallel - the number of filters one can use drastically reduces to 2 or less. Given that most modern soft synths have some serial or parallel filter solutions baked in for their users, Serum's 2 filter slots can be at least somewhat comparable to Zebra 3's 4 single filter. I'd definitely want (although I might not be using all the time) to see more filter slots and some pre-made multi filters in Zebra 3.

There are also some new things in the filter game in modern synths. Serum let you morph a filter, and both Phase Plant and Serum allow users to draw their own filters, or even import wavetables as filters. These are the things, while not super commonly used, open up a lot more possibilities for sound designers, which IMO worth considering to be added in Zebra 3.

Now let's talk a bit about routing and design. In highly modular synths like Massive X, you can route everything to everything else, sometimes even change the direction of signal paths and create feedback loops. The Zebra 3 routing system doesn't allow this level of flexibility. Given that the signal only flows down and the main OSCs work only in mixed mode, Zebra 3's design feels more like fixed architecture synths like Serum than modular. I hope there's a chance to allow more flexible routing, so that we can do things like cross FMO feedback or even feedback loops.

Now the part which I feel the most lacking: the modulator. While in Zebra 3 there are a lot of things one can modulate, and there are 12 built in modulators - 4 envelopes/4 LFOs/4 MSEGs (wouldn't directly count the mappers here, though I like the mapper design and find them useful in their own ways), the only thing featuring truly limitless modulation is the MSEG. The envelopes are shockingly (and, according to Urs, deliberately made to be) simple and does nothing much more than basic, linear AHDSR. The LFOs are also quite old school in design, and users can only choose from certain fixed waveforms. So anything remotely non-linear or more complex, say an non-linear envelope I'd use for physical modeling of a blown pipe or a one-time morphing/filter sweep with variable speed, requires an individual MSEG. Given that we only have a total of 4 MSEGs to use, I find myself running into the hard limit pretty quickly the moment I start to try to do something more complex. While I like the idea of old school envelope and LFO for quick things (and thankfully Urs mentioned that he can consider adding non linearity for the envelopes later), clearly the more standard approach prefered by the modern soft synths is to have everything built as single frame or table based MSEG. Serum, as a fixed-architecture synth, features 10 LFOs working as MSEG and 4 envelopes allowing non-linear behavior. Phase Plant allows as many LFO/MSEGs as users want. Even Massive X, quite simple in design, has 9 LFOs switchable between old school LFO mode and non-linear behavior. I see that we used to have 4 MSEGs in Zebra 2 and the number increased to 8 in Zebra HZ, so getting back to 4 in Zebra 3 feels like a retrogression.

That's the end of my review and thoughts for the beta testing. The experience kinda resembles my experience working with Massive X: great sounding synth, great osc and filter, a lot of sound design capability, but can occasionally feel frustratingly limited. Nevertheless, Massive X remains one of my favorite synths, although I might never see the day my complaints against it get resolved. And this again makes me grateful to hear that Urs isn't against the idea of adding more features and Zebra 3 will receive continuous improvements.

Anyway, even with its current form and limitations, I am sure Zebra 3 is still a very capable synth and people could do wonders with it. I have already purchased Zebra 3, and only time will tell how much it would get used in the future...

[Alexander_D]

Just interesting. How close U-HE to the official release of Zebra 3?.. So excited

[Urs]

@loctune

Thank you.

Yeah well, we limit some stuff and we have given the reasoning behind it.

As for drawable filters: Too much CPU for what they offer? - Happy to discuss later, don't have the time now to check this out in synth that have them. Note that our Oscillators *do* have drawable filters, at next to no extra CPU cost.

As for unidirectional signal routing: Yes, for audio. But Modulation can go anywhere you want, far beyond most competing products. And also, most competing products have no feedback loops, either. Zebra 3 is not about feedback loops but about combining different synthesis methods *without* the hassle of fully modular environments.

Simple LFOs/Envs: Complex ones are not suited for West Coast approaches. Zebra 3 has a lot of West Coast DNA. But happy to discuss improvements over time.

Regressing from 8 MSEGs in Zebra HZ back to 4 in Z3: The new MSEGs morph and cover more grounds than any preset using more than 4 MSEGs I have ever seen. Most of the time more than 4 was necessary for making up for inflexibility of non-morphable curves.

[johnnyberrygoode]

Ok, here's my super long review and thoughts after about a week beta testing this Zebra 3 revision. Zebra 3 is deep and I may have still been scratching the surface, but based on my personal experience, I think I've got a basic grasp of its functionalities and limitations. There are a number of things I really adore, but also a few things I feel lacking/lagging behind for a flagship synth released in 2026.

There will be mentioning of and comparison to a few other soft synths I own and use frequently in this post. I have no intention saying Zebra 3 as a product is inferior or superior to any of them. They are mentioned only so that I can talk about some comparisons and expectations. And while my post is not going to be all about singing praises, I have no intention to offend anyone nor to talk down the effort spent by the developer team in any way. Actually, I am grateful that I can beta test Zebra 3 and I would like to express my thanks to Urs for his help and his openness to communication. If anything I say here is wrong or inaccurate, please feel free to point them out.

I am happy to find Zebra 3 is not another Serum wannabe in 2026. Instead of another 2 osc wavetable & sample soft synth, Zebra 3 is something with its own design, thought and workflow. And I am glad to see Urs' attitude to define Zebra as a "pure synth" and focus on synthesis rather than get into samples. However, as Zebra's price is clearly flagship level, I also have quite high expectation about it: my common sense tells me that a flagship released in 2026 which refuses to do samples must have highly advanced pure synthesis functionalities. Turns out many of my expections are met, albeit not all of them.

Let's start with the main OSC. Zebra 3 took a 16 frame vector based approach instead of the standard 256 frame wavetable. I haven't done enough research or comparison to say which approach is better, but from a pure technical POV I can see more flexibility in the 256 frame wavetables (16x higher number of different waveshapes available in a wavetable and possibly better compatibility when importing larger wav files) while I think Zebra might have the edge on smoother morphing between different frames.

I have zero complain about Zebra 3's OSC sound. It's definitely on par with the best software synths in 2026 if not better. The additive mode is cool and the two OSC FX allow for quite some extra waveshaping functionalities (I'd like to have a feature request for an OSC FX which creates a sub-osc based on the current one, effectively a doubler with variable pitch). However, there are also a number of things I feel "incomplete", mainly the wavetable editor: there is no curve preset (coming next beta?), and there is no easy way to create a bunch of related frames in one pass. Users have to manually draw every frame in, which is inconvenient. The lack of spectral editing mode in Zebra 3's wavetable editor is also a huge concern. Given that two points with 0 distance degrades to a single one in the wavetable editor, there's no way to draw single vertical lines in the wavetable editor representing clean harmonics, and the produced waveform is always highly distorted in spectral mode. This lacking of a useable spectral editor (the mapper editor seems like a perfect fit here) makes its spectral rendering mode almost useless, at least in the current revision.

Then the FMO. During the beta testing, I keep feeling that the design of having individual FM oscillators is an afterthought remedy, trying to work around the issue of non-existing OSC FM (something still under evaluation according to Urs, really hope to see it soon) in Zebra. While it works and sounds good, it's still a heavily limited solution. Both the carrier and the built in modulator only do pure sine (distorted saw with feedback), which, since Zebra 3 isn't aiming to be a DX7 emulation, makes me feel kinda unsatisfied. Even the FM8 released 20 years ago does more waveforms than pure sine and has basic phase control, so for FM-specific osc in a flagship synth in 2026, I'd expect to see at least some more classical waveforms - or even better, editable single-frame wavetable - added as carrier and modulator options.

Now something super positive. The modal synthesis is what interests me most in Zebra 3. And it does such a nice job that I have literally nothing to complain here. The resonators are really nicely crafted. And I am happy that everything is plain CSV. Best modal synthesis I've seen in software synth.

There are a great number of analog-style filter options to choose from. And they all sound great. Yet there are only 4 single filters (some OSC FX can produce filter-like effect, but they aren't really filters) one can use in a single Zebra instance. Given that all the filter options are single filters, once you start trying something structrally more complex - let's say I am doing some MS-20 style HPF to LPF in serial or parallel - the number of filters one can use drastically reduces to 2 or less. Given that most modern soft synths have some serial or parallel filter solutions baked in for their users, Serum's 2 filter slots can be at least somewhat comparable to Zebra 3's 4 single filter. I'd definitely want (although I might not be using all the time) to see more filter slots and some pre-made multi filters in Zebra 3.

There are also some new things in the filter game in modern synths. Serum let you morph a filter, and both Phase Plant and Serum allow users to draw their own filters, or even import wavetables as filters. These are the things, while not super commonly used, open up a lot more possibilities for sound designers, which IMO worth considering to be added in Zebra 3.

Now let's talk a bit about routing and design. In highly modular synths like Massive X, you can route everything to everything else, sometimes even change the direction of signal paths and create feedback loops. The Zebra 3 routing system doesn't allow this level of flexibility. Given that the signal only flows down and the main OSCs work only in mixed mode, Zebra 3's design feels more like fixed architecture synths like Serum than modular. I hope there's a chance to allow more flexible routing, so that we can do things like cross FMO feedback or even feedback loops.

Now the part which I feel the most lacking: the modulator. While in Zebra 3 there are a lot of things one can modulate, and there are 12 built in modulators - 4 envelopes/4 LFOs/4 MSEGs (wouldn't directly count the mappers here, though I like the mapper design and find them useful in their own ways), the only thing featuring truly limitless modulation is the MSEG. The envelopes are shockingly (and, according to Urs, deliberately made to be) simple and does nothing much more than basic, linear AHDSR. The LFOs are also quite old school in design, and users can only choose from certain fixed waveforms. So anything remotely non-linear or more complex, say an non-linear envelope I'd use for physical modeling of a blown pipe or a one-time morphing/filter sweep with variable speed, requires an individual MSEG. Given that we only have a total of 4 MSEGs to use, I find myself running into the hard limit pretty quickly the moment I start to try to do something more complex. While I like the idea of old school envelope and LFO for quick things (and thankfully Urs mentioned that he can consider adding non linearity for the envelopes later), clearly the more standard approach prefered by the modern soft synths is to have everything built as single frame or table based MSEG. Serum, as a fixed-architecture synth, features 10 LFOs working as MSEG and 4 envelopes allowing non-linear behavior. Phase Plant allows as many LFO/MSEGs as users want. Even Massive X, quite simple in design, has 9 LFOs switchable between old school LFO mode and non-linear behavior. I see that we used to have 4 MSEGs in Zebra 2 and the number increased to 8 in Zebra HZ, so getting back to 4 in Zebra 3 feels like a retrogression.

That's the end of my review and thoughts for the beta testing. The experience kinda resembles my experience working with Massive X: great sounding synth, great osc and filter, a lot of sound design capability, but can occasionally feel frustratingly limited. Nevertheless, Massive X remains one of my favorite synths, although I might never see the day my complains against it get resolved. And this again makes me grateful to hear that Urs isn't against the idea of adding more features and Zebra 3 will receive continuous improvements.

Anyway, even with its current form and limitations, I am sure Zebra 3 is still a very capable synth and people could do wonders with it. I have already purchased Zebra 3, and only time will tell how much it would get used in the future...

An editable single-frame wavetable for fmo is a great idea.

[Urs]

Just interesting. How close U-HE to the official release of Zebra 3?.. So excited

Depends on whether people let me work or not...

[ThoughtExperiment]

Just interesting. How close U-HE to the official release of Zebra 3?.. So excited

Depends on whether people let me work or not...

[Alexander_D]

Just interesting. How close U-HE to the official release of Zebra 3?.. So excited

Depends on whether people let me work or not...

[loctune]

@loctune

Thank you.

Yeah well, we limit some stuff and we have given the reasoning behind it.

As for drawable filters: Too much CPU for what they offer? - Happy to discuss later, don't have the time now to check this out in synth that have them. Note that our Oscillators *do* have drawable filters, at next to no extra CPU cost.

As for unidirectional signal routing: Yes, for audio. But Modulation can go anywhere you want, far beyond most competing products. And also, most competing products have no feedback loops, either. Zebra 3 is not about feedback loops but about combining different synthesis methods *without* the hassle of fully modular environments.

Simple LFOs/Envs: Complex ones are not suited for West Coast approaches. Zebra 3 has a lot of West Coast DNA. But happy to discuss improvements over time.

Regressing from 8 MSEGs in Zebra HZ back to 4 in Z3: The new MSEGs morph and cover more grounds than any preset using more than 4 MSEGs I have ever seen. Most of the time more than 4 was necessary for making up for inflexibility of non-morphable curves.

Thanks for sharing your reasonings!

My argument regarding the MSEG though, is that even though the morphable MSEG is a beast, it takes yet another modulator to make the morphing happen. And within the current implementations, if more nonlinearity or user control is preferred, that second modulator to modulate the morphing MSEG has to take the form of yet another MSEG, albeit in singe frame…

Eventually, there needs to be something at the very bottom of the rabbit hole, in the form of single frame, non morphable MSEG style modulator to handle the base controls, implemented as envelopes or something else. And before that is done, I can foresee that quite a few of these morphable MSEGs will have to be used in the form of non morphable ones, thus the need of more MSEGs still feels somewhat valid.

Maybe some further things can also be implemented in the math module, which allows the use of a few easy shapes like linear/exponential/logarithmic and users can combine/loop them to do some simple modulations?

Anyway, all these can be looked into after the next beta and the official release. I do feel that a useable spectral editing mode is somewhat more urgent issue, and maybe worth addressing before the official release though…

Thanks again for your kind reply and look forward to seeing the next beta release!

[pdxindy]

I do feel that a useable spectral editing mode is somewhat more urgent issue

What do you mean by "spectral editing mode"?

[loctune]

I do feel that a useable spectral editing mode is somewhat more urgent issue

What do you mean by "spectral editing mode"?

There is a spectral curve mode in the main OSC, but I have not yet found any way of drawing a shape representing clean & simple harmonics in the wavetable editor.

An easy test is to try to draw a curve representing a clean sine wave - only a single fundamental or harmonic - which seemingly cannot be done in the wavetable editor. This requires two overlapping vertical lines, which merge into a single point in the current wavetable editor. One can draw curves with points with a distance, but that means potentially unwanted distortion…

Do you have a better suggestion to handle this?