TEST: VIRSYN CUBE, ADDITIVE SOFTWARE SYNTHESIZER
Additive goes PlugIn
October 14, 1999



Just so you know, you can also find a review of the VirSyn Cube 2.0 update on AMAZONA.de – CLICK HERE

ADVERTISEMENT
The VirSyn Cube from Germany brings additive synthesis to the virtual studio. Additive sounds like adding, and that's precisely what this unique synthesis method does when generating sound. Unlike subtractive synthesis, where something is subtracted from a complex sound using a filter, additive synthesis involves adding sine tones. The French physicist Jean Fourier (1768-1830) developed this method, and if you look at additive synthesis in a truly visionary way, it could recreate any sound found in the world simply by adding sine tones of different pitches and volumes – the dream of every synthesizer developer. So why doesn't this super-synthesizer exist after some 200 years? Well, quite...

Simply because accurately reproducing a sound requires a vast number of sine waves, each with its own separate volume, frequency, and panning envelope. Currently, no synthesizer DSP or computer CPU can handle this. An older instrument that utilized additive synthesis was the Hammond organ. More recently, the Kawai K5000 is a well-known example of this technology; it could generate a maximum of 128 partials and, due to its opaque nature, was hardly suitable for programming new sounds. Now, there's the VirSyn CUBE, an 8-part multitimbral software instrument with a remarkable 512 sine wave generators, designed to manage the parameter overload of additive synthesis. With the current version 1.5, additive resynthesis is making its debut on home computers, meaning that any sample can be imported and analyzed. CUBE generates the envelopes necessary for reproduction, specifying the level, panorama position, and frequency of each individual sine wave. We're excited!

Additive synthesis made easy
8-part multitimbral? 512 sine wave generators and 3 envelopes per sine wave? That would easily be 1536 envelopes per part! Help! But don't worry – a user-friendly interface was the top priority in the development of the CUBE. The envelopes of the individual sine wave generators don't need to be edited separately; instead, they are grouped into so-called macro parameters, meaning you can change hundreds of values with a single control.

The control of the sine tones and all available parameters is handled by the so-called "Sound Source." The CUBE features four of these (A, B, C, and D). Each Sound Source has its own Rate, Depth, Spread, and Ensemble controls. These macro parameters directly affect the 512 sine wave oscillators.

The Rate and Depth controls set the values for the pitch LFO's frequency and modulation depth. The Spread parameter adjusts the frequency intervals between the individual sine tones. Normally, in additive synthesis, the individual partials are in a 1:2:3:4 ratio, meaning the sound is perfectly harmonic. Turning the Spread control detunes the partials, making them inharmonic and therefore metallic. The Ensemble control adds warmth to the partials by slightly detuning them against each other, creating a warm, analog sound. Behind the scenes, this control manages no fewer than 512 LFOs. The Detune parameter detunes the fundamental pitch of the entire sound source. The corresponding control values are displayed when you hover your mouse over the control. To change the macro parameters of all four sound sources simultaneously, simply hold down the Ctrl key (Mac: Alt key) while adjusting the settings. Very handy!

Each sound source has predefined sources that make it easy to create new sounds. Custom sound sources can also be saved to the hard drive.



Advanced Additive Synthesis
In addition to the macro parameters, there are six partial tone parameters. The first 16 can be set individually, and the remaining 496 sine tones can be set in groups. This is done via graphical editors, and the values are entered using the mouse. The partial tone editor is accessed via the Edit button. Predefined envelopes are available via a context menu. The values entered here can be dynamically modified by envelopes, filters, or morphing. The following parameters are available:

• Partial: the basic volume of the individual sine tones
• Attack time of the volume envelope
• Decay time of the volume envelope
• Pan: Panorama position
• Filters with freely definable characteristics and special morphing properties
• Noise: adds noise with a freely definable spectrum to the partial tone

Global sound parameters

All four sound sources are affected by the global sound parameters Bright and Amount, as well as Phase and Partials. The Bright and Amount controls alter the sound characteristic between "bright" and "dark." In conventional synthesizers, this sound change is often simulated by modulating the cutoff frequency of a low-pass filter. In contrast, CUBE modulates the filter slope in the range of 0 to 30 dB/octave, as is the case with almost all acoustic instruments. This results in a significantly more realistic sound. The Brightness parameter can be linked to a modulation source (key track, velocity, modulation wheel, or aftertouch).

In the off position, the Phase switch allows all sine wave oscillators to oscillate freely, preventing an unnatural, sterile sound. For percussive sounds, however, the Phase switch should be activated so that all sine tones start synchronously with the keystroke.
The Partials parameter sets the maximum number of generated partials. And as with all plugins, CPU usage unfortunately increases with complexity. It's worth experimenting with this value. Higher values result in a better sound, especially in the low frequencies. However, if you don't have a powerful CPU, you should keep the Partials value as low as possible.

CUBE Specialty Part 1: Spectral Morphing
What looks more like a knitting pattern is actually one of the most important concepts in the CUBE. Using "Spectral Morphing," dynamic crossfading is possible between all macro parameters of the four sound sources. The morphing process is controlled by two morph envelopes (X-Morph, Y-Morph), each with up to 64 segments. The morphing process can also be controlled by modulation using MIDI controllers (X-Mod, Y-Mod).

Didn't understand? No problem. A short example will shed some light on the matter.



The envelope begins with the first segment (Attack) in the upper left corner of the 2D editor (yellow dot). The parameters of Sound Source A are also located in the upper left, and therefore, during the attack phase, the sound uses all parameter values of Sound Source A. Within a defined time (time values are not displayed in the 2D editor), the sound moves from Source A to Source B (upper right), meaning all parameters are continuously morphed from A to B. The next envelope segment on the Y-axis (Decay) runs from B to C. Finally, the next segment (Sustain) runs from C to D. After releasing the key, the sound morphs from D back to C (Release). However, this cannot be followed in the 2D editor because the two lines overlap. For better understanding, you should also refer to the familiar envelope view for the three possible envelopes (Volume, X-Morph, and Y-Morph).

The 2D editor is really fun! Even the smallest changes create completely new sounds. A fantastic tool for experimentation! What I'm trying to say is: you don't have to understand everything to create amazing sounds with the CUBE!

The Morph envelopes X and Y can be additionally controlled by the modulation sources pitch, velocity, key press, and modulation wheel. The envelopes can, of course, be synchronized to the MIDI tempo. With 64 segments per envelope, rhythmic sounds are also possible. New segments (points) can be easily added using the mouse and the Ctrl key. The time scale below the envelope display can be freely adjusted by simply dragging it with the mouse. Predefined envelopes can be loaded and custom creations saved via a context menu.

CUBE Specialty Part 2: Resynthesis
In the current CUBE version 1.5, VirSyn developer Harry Gohs has added a resynthesis module. Using the Import button, samples of any length can be loaded and analyzed. The sample is broken down into 512 partials, and the temporal evolution of level, frequency, and pan position is recorded and saved in a special file. The existing envelopes are replaced by those of the sample. Editing is still done via the macro parameters. While this function doesn't produce an exact copy of the original sound, the result is already very impressive! Natural sounds are particularly well-suited for the CUBE. Synthesizer sounds can also be imported. It's advisable to choose the lowest possible notes as the source. Fast rhythmic patterns still pose some challenges—but CUBE 2.0 will include several specialized import functions for different sound genres.

Three different analysis options are available, which must be selected before importing:

Simple
This option filters out unstable partial tones. This function removes unwanted artifacts and is particularly useful for simple, "electronic" sounds.

Blur
This function blurs the "contours" in the sound. Rapid changes in sound within the sample are significantly smoothed, resulting in a much softer sound.

Formant
When this option is enabled, the analysis attempts to extract the formant structure of the sample and uses it to create a model for the formant filter. Ideally, this avoids the notorious "Mickey Mouse" effect that occurs when transposing vowel sounds. However, this option is still in an experimental stage.

After the analysis, which can take anywhere from a few seconds to several minutes depending on the sample length, another envelope becomes available: the Time Warp envelope. This envelope, generated from the sample, allows the user, for example, to play the resynthesized sample backward or to speed up or slow down specific sections of the sample (time stretching). Just to reiterate: we're not dealing with samples here, but with sine wave oscillators!

If resynthesis doesn't sound good in the high-frequency range, you can easily remedy this by reducing the maximum number of partials. This is because, during resynthesis, the partials aren't arranged according to ascending frequency, but rather according to decreasing "importance" to the sound. When importing synthesizer sounds, this means that artifacts usually end up in the upper partials.
For most synthesizer sounds, 64-128 partials is a good compromise. Of course, with natural sounds, it's precisely these "less important" partials that contribute to the naturalness and appeal.

The Load function allows you to load over 100 pre-analyzed samples. These samples are excellent starting points for your own sound creations.



Effects
The CUBE effects section offers a global reverb effect and three part effects that can be set differently for each part.

The reverb section offers 24 preset programs, each with adjustable reverb parameters. The part effects include four different distortion and three delay programs (mono/stereo, 768 ms), as well as chorus, flanger, and phaser. The delay can be synchronized to MIDI clock, the internal arpeggiator tempo, or, in a VST instrument, to the sequencer tempo.

Sound library and sound
CUBE manages sounds in sound banks, which must be loaded individually into the CUBE. Each bank contains 64 sound programs. Eight sound banks, sorted by genre, are included with the CUBE. Registered users can download two additional banks from the VirSyn website.

A special highlight for all those who dislike tinkering is the slot machine! A single click on the one-armed bandit in the upper right corner of the screen is all it takes to create a new additive sound. And the results are truly impressive. Unlike typical randomize functions, this useful feature creates playable sounds!

The CUBE's sonic focus lies in dynamic pads and sequencer sounds, as well as lead and FX sounds. Its very fast envelopes allow for the creation of crisp, punchy sounds. The CUBE's sound programs are fresh and stand out from the usual subtractive synthesis fare. However, you have to appreciate additive synthesis. There are hardly any competitors. Only the two VirSyn synthesizers, TERA and D'cota, from the same manufacturer, use additive synthesis as a component.

Setup, arpeggiator and global settings
The 8 available parts can be conveniently configured in the setup and routed to the 8 possible output channels. Each part has a maximum of 32 voices, which can be limited using the "Voices" parameter. By using the same MIDI channel for different parts, multiple sounds can be layered. The Cube even has an integrated wave recorder. Simply click record and start playing.

The CUBE also features a simple arpeggiator (up, down, alt, random) – however, this is available separately for each part. In addition to all possible note values for the clock settings, there are also 8 rhythmic patterns to choose from.

Many parameters can be remotely controlled via external MIDI controllers. Assignment is done using the "MIDI learn" function: simply right-click the parameter to be controlled, select "MIDI learn" from the pop-up menu, and move the corresponding control on the external controller box.


CONCLUSION
Creating novel sounds with additive synthesis has never been easier. The user interface is exceptionally well-designed. It's clear that ease of use was a top priority during development. Pop-up menus and predefined templates for experimentation are readily available. This is why CUBE is particularly aimed at sound designers and sound engineers seeking fresh, original sounds. The integrated spectral morphing feature in CUBE is a real boon for additive synthesis. This function allows you to create impactful pads and constantly evolving FX sounds. As expected, the processing power consumption is quite high. For example, on a Pentium 4 with a 2.6 GHz processor (6 ms latency), a floating pad can consume up to 15% of the processing power for a single voice! The resynthesis already delivers astonishing results. Some tutorials with tips and advice on sound design would be welcome. My recommendation: definitely download and try out the CUBE demo before buying.

Plus

• Additive synthesis finally manageable
• very good user interface
• interesting, fresh sounds
• Spectral morphing between 4 sound sources
• German handbook

Minus

• High power consumption, only runs on CPUs with SSE/Altivec instruction set (Pentium III/IV, Athlon XP/MP, Apple G4/5)

Price

• €249

Left

• Manufacturer's product page for the VirSyn Cube

Source: /www.amazona.de/test-virsyn-cube-additiver-software-synthesizer/

 

Maybe a deadpan cross between farce, parody, critique and performance art...

In part inspired by 109 pages (and counting?) of one discussion-thread among others of Zebra 3 on KVR; the ideas of GAS vis-a-vis 'old thing' (emulations) revisits, the 'myth of progress' and 'all-knowing' AI; creative enshitification, online-forum hair-splitting/argument tangents; and a previous comparison hereon between two open source synths.

Or something like that. ;)

Check out Cameron's video I just posted in this thread. What is he trying to say, fundamentally?

 

Last edited: Nov 29, 2025 at 6:15 PM

...So which one? ^u' <-- right-side-up winky (old-school)

"Some users perceive certain u-he synthesizers, particularly Zebra 2, as sounding 'thin' or 'lifeless' when using default settings or stock presets. This perception is not universal but has been expressed in community discussions, especially among users transitioning from more character-driven synths like Native Instruments' Massive." ~ Brave search consolidated results quote

 

Last edited: Nov 29, 2025 at 6:09 PM

For the slow here...one is an outline of a zebra, the other a cube.

Viva la difference!!!

 

@Sackbut is stuck in a cosmic loop; currently, he's oscillating in an endless monologue universe.

 

So what are you saying? That the zebra is better than the cube because it's more complex? That it's just a one-trick cube? But look; the 'zebra' is flat and has no stripes, whereas the cube is 3D, so one more dimension. Next thing you're going to tell us is that it's not really 3D, but only a representation of 3D, right?

 

Last edited: Nov 29, 2025 at 6:37 PM

The Dark Zebra - Hans Zimmer and Howard Scarr

u-he Zebra Legacy is an amazing deal...

 

I might wait for Spline and Cadmium to show up on the sister site. Compare the two. Maybe ask Adam S. to consider creating an opensource bin for OsTIrus.

 

Last edited: Nov 29, 2025 at 6:48 PM

Bah. You have audiosex on the pillow right next to you. ;P

 

I wasn't saying that that the cube was inexperienced, it has made mint as a shape of the night and has had lots of tricks. Mine wasn't a post at all, but rather it was only a represention of a post!

I'm sorry that you disliked my post, I'll try harder the next.

 

Look again: I put an x, which is disagree. I liked it too, but the system only allows one per comment.

 

Verdict 2025

• Buy Zebra Legacy without hesitation if you want an unbelievably deep, great-sounding, future-proof semi-modular synth that will still be relevant in 2030.
• Only consider Cube 2 if you specifically need its particular granular/resynthesis engine for a very niche experimental project and you already own an older Intel Mac or Windows machine where it still runs decently.

Most people who owned both (including me) ended up uninstalling Cube 2 years ago. Zebra ate its lunch and then some.

So: Zebra2/Legacy is better – by a mile. Try the free Zebra Legacy demo or the free Zebralette 3 – you’ll hear the difference in 30 seconds.

 

Also known as Urs Allemann
and man.... please use ------>>

Spoiler: spoilers!

SPOILERS

 

Last edited: Nov 29, 2025 at 11:38 PM

Cube doesn't sound as good as Harmor, Loom, Razor etc. Comparing it to Zebra makes little sense, Virsyn Tera (a supersynth) is closer to its sound options since there are multiple synthesis methods in both of them. Anyway, it's hard to expect the greatest sound quality and sound design parameters when computer hardware was way less capable in the early 2000s, so it wouldn't support it. And seems Cube is basically not developed for a very long time.
Syne by Music Development supports thousands of partials, if you want a modern, clean additive synth with lots of control or something like Protean by Sonic lab, if you want crazy sci-fi texture sound effects etc.

 

Found a copy of the image previously-mentioned, but mine replaced 'message' with a different word (and had no coloured text).
(edit) Oh hey, you can even spoiler an image...

Spoiler

What do you make of this from Urs (I think it's from KVR):

What does that mean, 'turn samples into Curves' or resynthesis tools 'are usually sample based'?

Is Zebra 3 an (in part) additive synth or what? What's a 'Curve' in this context ('waveform'?) and what happened to the sample? How is it being interpreted/reinterpreted/manifest? Where are the partials? Are they sine-wave partials? Or granular clippings?

Maybe I should see about asking Urs.

If synths are going to have multiple synthesis under one roof, it would still seem fair to compare one kind to another synth that specializes in that kind, yes?, especially if one or more of the synthesis methods in the multiple synths are somehow less capable?

Fair enough. Perhaps Zebra 3 will be some people's new 'desert-island synth'. BTW, aren't islands less desert and more moist and tropical?

Nice edit, mod, thanks! I'll keep the consolidated responses format in mind for the future.
(Edit; I adjusted your attachment size in the spoiler before the attachment link below was evident. Unsure how to resize it back down. Anyway, it seems to look ok on my end.)

 

Last edited: Nov 30, 2025 at 4:02 PM

1) Will be probably explained in the user manual. I don't have the time to read kvraudio threads about how it works, but it is probably already explained. There is also the new Zebralette free (I haven't downloaded it yet), which has one of the additive oscillators of the new Zebra, you can inspect it on your own and find on your own what it does. I have Zebra 2 only, which is obviously slightly different. I think Urs intentionally didn't allow resynthesis (yet), because probably doesn't want it be marketed toward only film composers/ambient crowd etc since that's the fate of such synths most of the time (Absynth, Alchemy, Omnisphere etc). People playback or mangle samples, don't explore the deep synthesis features. Look at even Kontakt - it is probably a more capable synth than old Massive for many years since they added synthesis, but is mainly used to playtback samples (even synth samples, xd, which is even funnier).

2) Probably not a fair comparison - generalist do-it-all instruments offer less control and have less depth. Stuff like Reaktor and MaxMSP is an exception, but they are way too complicated for average synth enthusiast and the CPU hit is not worth it most of the time, so few people in the whole world use such complex programs.

 

Good points above.

Yup

Yes, probably Plugdata too.

 

Desert...as in deserted...and not like dessert.

p

 

That looks moist and tropical and something one could order at Dessert Island Cafe. They could add a tiny edible sugar figurine on top, playing a synth. With live music played on various desert island synths on Friday and Saturday nights.

I think I like the idea of creating crappy music with Cube 2 while 'everyone else' creates fantastic music with Zebra 3.

I could show up some nights with my Cube 2 and something like this:

Spoiler

Best performance gets a free dessert in a doggy bag after the show.
Now all I have to do is figure out some kind of outfit.

 

Last edited: Nov 30, 2025 at 11:25 PM

The big takeaway for me is not comparing anything at all. Find what you like that suits your workflow and musical visions and make music with it. It doesn't matter what any review or any person says about what you like. You like it or you do not.
The only question for me if I do not already know the answer, is "Which synthesizers are capable of doing this....". I'll then go out and try the suggestions and settle on one I like, regardless of opinions or reviews. i will also be looking for myself as well. Why bother comparing the ridiculous number of synthesisers - Dune, Zebra, Serum, REFX Nexus, Arturia, N.I etc etc etc ???? You like whatever you like and for me that is exactly how it is supposed to be.