Zebra 3 Public Beta 2 Revision 20552

[tasmaniandevil]

* Browser: Double-clicking a preset now switches to the Synth tab: Not happening (W11)

I tried it in Reaper under Windows 11, and it works without issues. A simple double-click on a preset closes the preset browser and switches to the Synth page. No modifier keys are needed.

Which host (DAW) are you using?

[ThoughtExperiment]

BTW, I checked Reaper and I can't find any way to map any CC to channel pressure - I guess because AT isn't a continuous controller - it's easy enough to map one CC to another CC, but not aftertouch. I've posted a question on the Reaper forum just in case, but I probably already went through all this years ago. If I ran out of mod sources like this in Zebra2 my workaround was to use the XY pads instead, but it was the only time I used them...

AT is a continuous controller.

There are free plugins that can convert CC's to AT.
PizMIDI can do this (midiConverter3)
PlugInGuru ModMate also can do this.

There are also JSFX Scripts for REAPER out there that users share. I don't use Reaper (since many years). It's easy to convert in Bitwig so I don't use those plugins, but they should still work in Reaper.

Sorry, I overlooked this post last night. I've had a few responses on the Reaper forum, one of which is a script which I'll try out later, since that would be the neatest fix.

Thanks again for your help, and particularly for understanding that this is actually a problem. To be honest I'm a little disappointed with the response from uHe on this issue. Maybe I'm not shouting loud enough

[Ghul]

Hi,

while experimenting with Delta-X for bass sounds, I noticed that faster rhythmic LFO modulation using smooth, sinusoidal waveforms (sine / triangle) tends to become aggressive and a bit rough quite quickly (for example at dotted 8ths or dotted quarters).

For clarity: in this setup I’m only using sinusoidal modulation shapes, and I’m using the curves as as modulator.

In comparison, using a Zebra oscillator as a modulator for the FM oscillator stays noticeably smoother and more controlled at similar rhythmic rates.

I was wondering whether this difference mainly comes from the technical nature of Delta-X rather than from the modulation itself, or if I might be overlooking something in my setup.

Just curious whether this observation lines up with what’s technically going on under the hood.

[ThoughtExperiment]

This is a hack but it works. Save this as a text file in the Reaper effects folder, changing the 'CCno = 16' part to the CC number you want to convert to AT

Code: Select all

desc:CC#ToChannelPressure

@init
CCno = 16; // change me to change the CC#

@block
while (midirecv(offset,msg1,msg2,msg3)) (
  (msg1 & 0xF0) == 0xB0 && msg2 == CCno ? (
    msg1 = 0xD0 + (msg1 & 0xF);
    midisend(offset,msg1,msg3);
  ) : (
    midisend(offset, msg1, msg2, msg3);
  );
);

You can then add a JSFX plugin CC#ToChannelPressure as the first plugin in the chain, before Zebra3.

I'm afraid I've no idea where I got this from so can't give proper credit...

Sorry for the late reply but I overlooked this last night. Thank you so much for this - I just tried it out and it works a treat You're right, it is a hack but it's only a small hack and it works

[Urs]

Hi,

while experimenting with Delta-X for bass sounds, I noticed that faster rhythmic LFO modulation using smooth, sinusoidal waveforms (sine / triangle) tends to become aggressive and a bit rough quite quickly (for example at dotted 8ths or dotted quarters).

For clarity: in this setup I’m only using sinusoidal modulation shapes, and I’m using the curves as as modulator.

In comparison, using a Zebra oscillator as a modulator for the FM oscillator stays noticeably smoother and more controlled at similar rhythmic rates.

I was wondering whether this difference mainly comes from the technical nature of Delta-X rather than from the modulation itself, or if I might be overlooking something in my setup.

Just curious whether this observation lines up with what’s technically going on under the hood.

I guess this might be related to the resolution parameter. The rendering process in the oscillator is pretty involved, so we can't magically update everything on a sample-per-sample basis. Instead, we update everything 200, 800 or 2000 times a second. This is also when external modulation is evaluated. Hence, when the modulation gets fast, it is less smooth than, say, and FMO which is designed for audio rate modulation.

There are pros and cons with every technology. Zebra's oscillators are extremely versatile for sound generation, but the resolution vs CPU ratio is a trade off.

[Ghul]

Hi,

while experimenting with Delta-X for bass sounds, I noticed that faster rhythmic LFO modulation using smooth, sinusoidal waveforms (sine / triangle) tends to become aggressive and a bit rough quite quickly (for example at dotted 8ths or dotted quarters).

For clarity: in this setup I’m only using sinusoidal modulation shapes, and I’m using the curves as as modulator.

In comparison, using a Zebra oscillator as a modulator for the FM oscillator stays noticeably smoother and more controlled at similar rhythmic rates.

I was wondering whether this difference mainly comes from the technical nature of Delta-X rather than from the modulation itself, or if I might be overlooking something in my setup.

Just curious whether this observation lines up with what’s technically going on under the hood.

I guess this might be related to the resolution parameter. The rendering process in the oscillator is pretty involved, so we can't magically update everything on a sample-per-sample basis. Instead, we update everything 200, 800 or 2000 times a second. This is also when external modulation is evaluated. Hence, when the modulation gets fast, it is less smooth than, say, and FMO which is designed for audio rate modulation.

There are pros and cons with every technology. Zebra's oscillators are extremely versatile for sound generation, but the resolution vs CPU ratio is a trade off.

Thanks a lot for the technical explanation — that really clears things up.

I also just wanted to say how incredibly strong the oscillator is overall. It genuinely feels like a sound synthesis toolbox on its own. Exploring the oscillator FX in combination with the spline editor is extremely fun and inspiring — to the point where I sometimes catch myself wishing for a third oscillator FX slot

It’s honestly impressive how far you can go with just a single oscillator. Creating nearly finished, complex, growly / neuro-style bass sounds — sounds that often require multiple oscillators and maybe some effects in other synths — is really something special.

Great work, and thanks again for being so open about the technical details.

Cheers

[charlee8t]

I don't understand, what the LFO free trigger option really does. Shouldn't a LFO with free trigger setting then not reset or change the phase, but running as a continuous curve?

Here there are sometimes phase shifts, when playing a new note. Global is set to legato.

I was wondering the same as well. Bumping this question for visibility

[Urs]

I don't understand, what the LFO free trigger option really does. Shouldn't a LFO with free trigger setting then not reset or change the phase, but running as a continuous curve?

Here there are sometimes phase shifts, when playing a new note. Global is set to legato.

I was wondering the same as well. Bumping this question for visibility

We have changed something in the LFO a few days ago, right before the current beta release. I do not know exactly what the change was.

Note that when Reallocate is turned off, the voice manager round robins through all 16 voices. There might be a bug where it even does that when Reallocate is turned on. So free running LFO phase may change from voice to voice. Not sure.

Revisiting the voice management behaviour is one of our last bugfix tickets before the final release. So there's a good chance that the perceived LFO reset behaviour is simply a bug in voice management.

There's also a good chance that, given there were open bug tickets, the documentation is missing or based on outdated information in the user guide.

[charlee8t]

I don't understand, what the LFO free trigger option really does. Shouldn't a LFO with free trigger setting then not reset or change the phase, but running as a continuous curve?

Here there are sometimes phase shifts, when playing a new note. Global is set to legato.

I was wondering the same as well. Bumping this question for visibility

We have changed something in the LFO a few days ago, right before the current beta release. I do not know exactly what the change was.

Note that when Reallocate is turned off, the voice manager round robins through all 16 voices. There might be a bug where it even does that when Reallocate is turned on. So free running LFO phase may change from voice to voice. Not sure.

Revisiting the voice management behaviour is one of our last bugfix tickets before the final release. So there's a good chance that the perceived LFO reset behaviour is simply a bug in voice management.

There's also a good chance that, given there were open bug tickets, the documentation is missing or based on outdated information in the user guide.

Thank you for the prompt reply!

[liquidsound]

* Browser: Double-clicking a preset now switches to the Synth tab: Not happening (W11)

I tried it in Reaper under Windows 11, and it works without issues. A simple double-click on a preset closes the preset browser and switches to the Synth page. No modifier keys are needed.

Which host (DAW) are you using?

Thanks for answering.
I tried Bitwig 5.x, Reaper 7x, Reason 13.5, FL Studio Latest.

Unusual...

EDIT: Maybe this is the wrong thread. I should post in the Bug Report thread.
I may try later over there.

[Sampleconstruct]

I think something is wrong with the mixer module, input 2 is much softer (like -10 dB) compared to the original volume in the lane itself, input 1 is fine.

[SamDi]

We have changed something in the LFO a few days ago, right before the current beta release. I do not know exactly what the change was.

The issue I observed, was there in the previous version as well.

Note that when Reallocate is turned off, the voice manager round robins through all 16 voices.

Reallocate is turned on. The phase jump happens with the same note played.

But when it is turned on - isn't per specification the phase different for different voices nevertheless? What's the sense to have free running LFOs, different for each voice, even if you don't want it? How do I get in Zebra a free running LFO, globally the same for all voices, e.g. to apply a filter sweep for each voice the same without getting retriggered at all?

[magickalmutagen]

Would you consider adding a way to invert a modulation source (perhaps via a curve) before it reaches its destination? It's probably already possible but I couldn't figure it out.

...

So for Velocity and envelopes specifically you'd simply map one of Ctrl A..D to the Velocity slider of an envelope, as in, keep the Velocity slider at 0.00, but then map e.g. Ctrl A onto that with a depth of +100.

Useful and works how I was hoping my macro would work, but like you said, it's only for the envelopes.

For anything modulated through the Mod Matrix, you have multiple choices. The first one that comes to mind is use Velocity as "Via" Source, and use Ctrl A..D to modulate that Via Depth from 0 (no Velocity) to 100 (full Velocity). But that wastes a Modmatrix slot for each modulation.

So that does achieve the desired result, but on top of sacrificing a mod matrix slot, it seems I need to manage the state of two controls: I must ensure that the 1st slot's source is max value else the 2nd slot which modulates the 1st's via depth won't behave as expected.

Another option is to setup a ModMath module to crossfade between Constant and Velocity as your Via Source, and use a single Matrix slot to map Ctrl A..D onto the ModMath X value. However, "Constant" is not available yet (should be very soon), so you'd temporarily use a second ModMath just for this. I can make a screenshot if needed, once I'm in the office later today.

This sounds like it could be the answer, although it's certainly more complex than a toggle or curve to be able to invert the source modulation which I believe is all that would be needed. I think I managed to try this suggestion out successfully, but for some reason the velocity range is much more limited (doesn't get as quiet) compared to when velocity is assigned directly, so it's still not quite right.

This is an interesting problem that In hadn't thought of before. There's no way currently to do this globally for Velocity itself or for its special role in envelopes.

I'm a bit confused by this so I'm not sure if that's really the problem (at least in my case). Being able to affect its sensitivity globally could have some benefits, but more generally I quite often find changing the curve of a source from linear to something else in the matrix a nice way to add further interest without needing to set up an additional source Osc / MSEG etc, and, again, would be the simpler solution.

[MCMorrise]

This sounds like it could be the answer, although it's certainly more complex than a toggle or curve to be able to invert the source modulation which I believe is all that would be needed. I think I managed to try this suggestion out successfully, but for some reason the velocity range is much more limited (doesn't get as quiet) compared to when velocity is assigned directly, so it's still not quite right.
…
more generally I quite often find changing the curve of a source from linear to something else in the matrix a nice way to add further interest without needing to set up an additional source Osc / MSEG etc, and, again, would be the simpler solution.

I tend to agree. We don't need a whole modulation remap feature like in Serum 2, but I think most users would concur that an "invert" toggle and a basic log/lin/exp curve thing would be rad. Of course, if that's too much to ask I understand. A lot of work has gone into this already and it's really quite phenomenal what Urs et al. have done and how much we can accomplish with the tools on hand.

[exmatproton]

feature request; more intermittent steps for the mappers. Tbh, i would just want to see all steps being available from 1 to 128