Zebra 3 Beta Bug Reports

[Viktor [TUC]]

- Set the trigger to "Sync" (should be default)
- Set the waveform to "Rand Glide" ("Rand Hold" seems to have the same problem)
...
= It appears that the LFO moves to a random value at the start of every note, contrary to what I'd expect from the Sync trigger mode.

That's what random means though, there's no fixed amplitude value to reset to. These options in the "Trigger" menu are all about phase positions. Can you describe the LFO behaviour that you'd expect to hear with the random modes? You press a note, and the LFO (set to Rand Glide/Hold and Sync) should start with the same value, then go into random amplitudes?

EDIT: Under the Delay Time knob in the LFO, select "Dly+Fade" (when you use Rand Glide) and turn up the amount a bit, that should pretty much do what you want. LFO starts at zero, reliably, and the random values then fade in.

Viktor
u-he team

[NAD]

Can you describe the LFO behaviour that you'd expect to hear with the random modes?

Thanks for responding, Viktor. What I'm expecting is some way to achieve the same effect as the LFOG in Zebra2, i.e. an LFO that basically ignores note-ons such that you can play overlapping notes without the LFO jumping to a new value on each note.
As an example, try this in Zebra2 to see what I mean:
- Start with the init patch
- Set a very long release in ENV 1
- In the VoiceMix Volume1 direct modulation slot select "LfoG1" and turn it all the way up
- Now in LFOG 1 select the "rand glide" waveform and dial in a fairly slow rate
- Play some notes and observe how the lfo doesn't react to individual note-ons
I expected to achieve this in Zebra3 by choosing the "sync" lfo trigger mode, and for the non-random waveforms it does behave how I would expect. However if [in Z3] you try the same thing I outlined [for Z2], you'll get jumps for each note-on. Does that make sense?
(On second thought, the effect might be more apparent if you modulate the oscillator pitch rather than volume.)

[gassle]

- Global Voice count doesn't seem to change, unless I switch it to Mono and back to Poly, then it updates. (at least with some initial patches I created). Seems to work ok after I do this switch once in a patch.

Voices are usually cycled through all 16, with few exceptions. The global voice setting just sets the maximum number of voices that are heard at the same time. If this number is exceeded, then voices are "discharged", i.e. quickly faded out, to bring it back to the desired maximum. So for a brief time, the number of voices can go higher.

Thanks for clarification! The voices indeed clear after envelope tails end.

[Urs]

There is a good chance that some settings mess things up, e.g. I think there's a bug with Release as trigger for envelopes, where voices never stop processing

[Sebastriani]

(EDIT: Just saw this is already reported, sorry for the noise...)

Another thing: I have mapped controller knobs to CC 2,11,75 and 76, so that they can control the "Ctrl A-D" knobs. This works, however, the knobs are not visually moving alongside inside the VST (they do for MW and PB, though). Again: the CC reaches the VST just fine and modulates the sound accordingly, just visually the knobs stay at zero.

I'm running Zebra 3 in Reaper on Debian 13, I cannot make Control A to respond at all, I mean not just that they don't move but more importatnly they do not modify the value of the control assigned to them, neither by sending CC2 or assigning it another CC in the Global or Per Instance config. Control B works in some presets, Control C and D I think work in all presets. Of course I'm talking about presets that have those Controls assigned.

EDIT: Forget it, it was because I had MPE enabled in my Korg Keystage.

[brightonrock]

If the 'full' oscillator editor is on screen when you 'init' a new patch, you lose the source and render options from the bottom half of the screen on init (the bottom of the screen switches to show the mini view of the oscillator)

[exmatproton]

When using 'spectral decay' as a modifier, retriggering seems to be broken when you use 16 voices. The decay isn't being triggered when all voices have been used. At 15 polyphonic voices or less, this isn't happening. Actually, at 15 voices or less it happens as well, but less often. Same issue when selecting MONO + Overlap MONO

any word on this behavior?

[Urs]

When using 'spectral decay' as a modifier, retriggering seems to be broken when you use 16 voices. The decay isn't being triggered when all voices have been used. At 15 polyphonic voices or less, this isn't happening. Actually, at 15 voices or less it happens as well, but less often. Same issue when selecting MONO + Overlap MONO

any word on this behavior?

I wrote about it. It is probably related to Voice Stealing, and we have no good solution for it yet.

[Teoman I]

I've noticed this shut down/crash behaviour.
When the filter palette is open, pressing 'Esc' key a few times closes Zebra 3 down.
Doesn't seem to mater what filter is selected.
(Windows 11 Pro, Reaper v7.56)

[PieBerger]

I wrote about it. It is probably related to Voice Stealing, and we have no good solution for it yet.

"Let's go to the Winchester, have a nice cold pint, and wait for this all to blow over."

[gassle]

I have to say playing with Zebra3 for more than a week, this is taking the crown for me from the Korg Multipoly for being the most analog sounding instrument plugin. Sounds gorgeous no matter where I push it, as long as you are careful about details like matching the routings properly for synths you wanna emulate, e.g. modulating filter fm with a single osc instead of the whole stack, using folders/distortion polyphonicaly in correct places etc.

Only trouble I'm having is the occasional crashes which almost always occur while opening or closing the plugin GUI while using AU version in Ableton Live 12, on M1 Sequoia. And that 1-2 seconds wait for opening/closing the GUI and sometimes audio also mutes for a sec while doing that which I don't observe with other plugins. (*Not opening a new instance, but while opening and hiding the gui)

[the3ug3reeder]

Zebra 3 Version: Public Beta (December 2025 Revision)

Description:
When attempting to restrict the **Discover** smart folder to only show presets from the **User** directory, the browser fails to maintain the path restriction. Even when the **User** path is successfully locked (appearing below the Search field), selecting the Discover category causes the browser to pull random presets from the global library (including the Local/Showcase folders) rather than drawing exclusively from the restricted User path.

Steps to Reproduce:
1. Open the Preset Browser.
2. Double-click the User folder in the directory panel to restrict the search path.
3. Verify that **User** appears in the path bar below the Search field.
4. Click on the **Discover** smart folder.
5. Observe that the resulting presets are pulled from folders outside of the restricted User path.

Expected Result:
The Discover folder should act as a dynamic query restricted to the currently selected path, providing a random selection of presets drawn *only* from the User directory.

Actual Result:
The Discover folder performs a global query, ignoring the active path restriction.

[bhinton]

I think this would be considered a bug:
Exciter - once max polyphony is reached (eg via a fast trill with sustain down) it stops triggering the modal module at all (or simply loses its attack - hard to tell).
As opposed to using a noise module with a fast envelope to excite the modal module, in which case polyphony behaves as expected (so it's the exciter not the modal).

[Sampleconstruct]

I'm not sure if this has been mentioned: Controls A-D don't visually reflect their saved positions when reloading a patch, the saved values are correct but the knobs are all turned hard left.

[Urs]

I think this would be considered a bug:
Exciter - once max polyphony is reached (eg via a fast trill with sustain down) it stops triggering the modal module at all (or simply loses its attack - hard to tell).
As opposed to using a noise module with a fast envelope to excite the modal module, in which case polyphony behaves as expected (so it's the exciter not the modal).

Yep, that is a bug. It is the result of our internal struggle on whether to keep "Trigger" or "Gate" as trigger sources or not, and "Gate" won, but now "Trigger" is sorely missing.

The actual issue is that when voices are stolen as max polyphony is reached, the stolen voice gets retriggered with the new note. But as Gate may still be high on that voice, the Exciter is not retriggered, unlike if it was triggered by the Trigger signal.

We will fix that, somehow, I have yet to figure out how.

Best solution for now is to set Poly to 8-12, so that enough voices are "discharged", and "Discharge" to "Fast" so it's quick enough, in order to avoid note stealing.