Official Serum 2 thread!

[Igro]

- Added Voice Steal Retrigger options to Envelope graph right-click menu


Guys, I'm so grateful that you aded this feature! I can't believe my eyes. Thank you!

Can you explain why, I don't know what I don't know

I will paste Olicash response to a similar question on discord:

"It determines the level from which the envelope starts when a voice is stolen (this includes playing a second note when one is already playing if mono enabled). Default settings are that the amp env (env 1) starts from whatever level it was at in the previous voice before it was stolen, as this helps reduce clicks. Envs 2-4 start from whatever the start level is set to (default 0). Note the amp env (1) does not have an editable start value, it always starts from 0."

[mabian]

I'd still like to see a linux native version....

Totally, it doesn't work at all in Wine currently...

- Mario

[Morty-C-137]

Don't know if it was there before, but holding down option (alt) + shift + left mouse click now toggles between unipolar an bipolar mode for the modulation.

[phreaque]

Don't know if it was there before, but holding down option (alt) + shift + left mouse click now toggles between unipolar an bipolar mode for the modulation.

Yes it was already there since version 1

[Razzia]

Considering what Serum 2 is doing under the hood (audio rate modulation for almost everything) I'd be really surprised if the CPU usage gets any lower than it is now.

A tale of two synths
People for whom it's working as expected see all the comments about performance and say "well it's got a lot of horsepower, of course it's going to be CPU hungry"

People with systems that Serum 2 doesn't like (particularly AMD it seems) run a basic monophonic arp patch with no especially taxing processing or audio rate modulation and get 25-45% CPU hit.

It does seem to have improved somewhat over the last couple updates but it eats literally 5x the CPU of other synths with similar, notoriously CPU hungry functions, like Phase Plant or Avenger.

[tumface]

People with systems that Serum 2 doesn't like (particularly AMD it seems) run a basic monophonic arp patch with no especially taxing processing or audio rate modulation and get 25-45% CPU hit.

Basic monophonic arp on my 2017 AMD Ryzen 1 system:

FL64_D3Is5RVtpd.png

But I do agree it uses more CPU on this system than Serum 1 did. 2.0.20 improved it substantially over how it was before, but it was usable before, and is very usable now. On my M1 Mac, there is no difference with Serum 1 (and sometimes Serum 2 is more efficient.)

It does seem to have improved somewhat over the last couple updates but it eats literally 5x the CPU of other synths with similar, notoriously CPU hungry functions, like Phase Plant or Avenger.

I never thought of Phase Plant as being particularly CPU hungry. Its basic oscs and (kinda crappy) filters, even with audio rate modulation, don't seem much higher than anything else. Phase Plant *can* use a lot of CPU, if you build a patch with lots of polyphonic effects and other stuff. But what would you want them to do? Make Phase Plant lock itself out if you try to add too much stuff to a patch?

I've never used Avenger, so I have no clue about that one.

If I want to make or use a patch with 90 oscillators with stacked unison across multiple octaves, I'm going to pull out Hive, Dune, or Icarus. It was the same during the Serum 1 days. Low CPU usage under high voice count has never been something I associated with Serum.

I do think that a lot of the Serum 2 factory presets use a lot of CPU for the sounds they're synthesizing. A lot of the presets use arpeggiators with high voice count limits and long release tail timings, plus stacked copies of FX in the FX rack. It doesn't leave a great first impression if you're on an older system and stepping through the factory presets. The patches I make for myself aren't like this at all.

[tumface]

Ah, one more thing that might be relevant. You might think that a basic patch like that isn't using audio rate modulation. But, it is: the envelopes.

Here's the output of Serum 2 vs Hive 2, with both having a very short amp envelope and no filter, playing a DC waveform. (Left, Serum 2. Right, Hive 2.)

FL64_Eg4nhICnu9.png

In addition to being able to customize the curve, which I've taken advantage of in this example, you can also see that there aren't any interpolated line segments in the Serum 2 output. Are you going to hear this in a preset? Well, yeah, if it's the right kind of preset. And also, no, you won't, if it's another kind of preset. This isn't hating on Hive 2. It's one of my most used synths, and I love it. Different goals, different tools.

(In Serum 2, I created a DC waveform. In Hive 2, I did this by playing a square wave at a really low frequency, and making sure the phase didn't cross the halfway point. The little artifact at the start in Hive 2 always seems to be there, even when using a sine wave or other waveform.)

[Alexander_D]

I love U-HE. Especially Diva, Zebra and Repro. But regarding Hive I prefer Serum 2 for sure. And of course it's an amazing hybrid synth. I am sure, that in the next updates it will be better and better.

[Morty-C-137]

Don't know if it was there before, but holding down option (alt) + shift + left mouse click now toggles between unipolar an bipolar mode for the modulation.

Yes it was already there since version 1

Anyways, glad to know it is there. Was missing easy access to this from being used to Vital.

[djanthonyw]

Still not sure what the appeal to Serum is. The Sound isn’t special and the CPU as mentioned above is horrible. If any plugin uses that much CPU the sound better be top notch to justify it. There are plenty of other options that sound better and use hardly any CPU. I guess Serum just became popular fron good marketing and large artists using free copies for Xfer’s marketing campaigns.

[tumface]

I think it's the all-around best sounding wavetable synthesizer that exists. And the UI is great. I have used many, many synthesizers.

But I don't use it exclusively. I have a bunch, and I love them all.

[WasteLand]

the appeal is the spectral oscillator.
cpu usage isn't bad, not bad at all.
also the new non-linear filters..
wavetables, yes of course, mainly i use them as modulators....
the granular oscillator isn't bad, usefull even.

with all the cross mods... possible ánd more... well create new sounds from own samples... ot is quite a pleasure ánd it ends up in tracks.

[swilow11]

Serum doesn't really have a sound. It's basically popular because it's super versatile and easy enough to use but with heaps of depth if you want to push things. I don't think I've ever seen Serum being "marketed". Personally, it is the synth I use for maybe 70% of what I do. I've sold so many others like Spire, Hive, Current, Pigments, Falcon, all of which are great but all of which can be done easier and often better by Serum.

What synth can you do audio rate modulation on lfo nodes? Not sure I even want to do this but I at least can...!

FL64_Eg4nhICnu9.png


In addition to being able to customize the curve, which I've taken advantage of in this example, you can also see that there aren't any interpolated line segments in the Serum 2 output. Are you going to hear this in a preset? Well, yeah, if it's the right kind of preset. And also, no, you won't, if it's another kind of preset. This isn't hating on Hive 2. It's one of my most used synths, and I love it. Different goals, different tools.

(In Serum 2, I created a DC waveform. In Hive 2, I did this by playing a square wave at a really low frequency, and making sure the phase didn't cross the halfway point. The little artifact at the start in Hive 2 always seems to be there, even when using a sine wave or other waveform.)

What do you mean by "interpolated line segments"?

Interesting post BTW.

[tumface]

What do you mean by "interpolated line segments"?

The modulation signals in Hive are calculated in steps and then interpolated together. If you look closely at the output, you can occasionally see it in certain cases where drastic changes in the control signal occur quickly:

firefox_TZ8nMt6IKa.png

This is a totally normal thing to do in DSP and 99.99% of stuff does it. You probably already know this as being "not audio rate modulation." It saves a ton of CPU, and when done properly, doesn't introduce any aliasing or negative audible artifacts. It can, in some cases, make envelopes or LFOs feel a bit different. Usually, you won't be able to find any difference unless the envelope or LFO is running very quickly.

[Gam456]

Serum doesn't really have a sound.

A synth can be versatile and have his own character.
That's why I'm not playing with Serum, no personality at all. I get bored.