My only suggestion is the some tweaks in synth engine to have ability to make it sounding like classic VA synths - Norld Lead, An1x, Virus etc. I don't know what is diffirence in DSP, but current softsynths sounds nothing close to them, it is almost impossible to get that "classic" sound with any modern vst synth... I think the main difference is in the high frequency domain - hardware va-s oscillators spectrum frequency response closer to white noise distribution of harmonics with a fully open filter, that gives brighter and cleaner sound in the mix overall, (most softsynths tend to more like pink-noise response curve), and maybe give option to turn on some aliasing to add that metallic "BUZZ" on high frequencies, this is most noticeable in that classic hw VA supersaw-type sounds.
And ability to keytrack that aliasing - more on very low bass notes (to make attacks and overall highs brither), and almost non-existent on high notes (when it much more noticeable and messing with actual harmonics of sound). So...my main idea is: almost all supersaws made with current soft-synths sounding dull and noisy (mushy - just pink noise instead of clean harmonics) in the high frequency domain, don't know what causes this... Maybe too much bandlimiting and lack of oversampling?
And one more thing that very important to me - oversampled wavetables (192 khz or more)
On the conventional wavetables that everyone uses today low octave notes cuts too early in the spectrum (for example sawtooth). This is VERY wrong. Spectrum should be full to 20 khz, independent from how low your note is.
And want feature of oversampling everything, i like crisp clean sound with audio-rate modulations like PM/FM etc.
And multicore support!
Zebra 3 feature suggestions
-
- KVRAF
- 6780 posts since 17 Dec, 2009
Uh...
diva already has all of that - including optional aliasing on the digital OSCs.
Hive also has a full spectrum without a rolloff up to nyquist. (Well its a bit more complicated but generally its true)
If you oversample you need to use LP filters, which either add latency or change phase response (== waveform shape)
Having full bandwidth without aliasing calls for a different method. Urs discussed it re: hive already
I like how zebra sounds; but you’re really barking up the wrong u-he tree imo.
Have you tried Hive?
diva already has all of that - including optional aliasing on the digital OSCs.
Hive also has a full spectrum without a rolloff up to nyquist. (Well its a bit more complicated but generally its true)
If you oversample you need to use LP filters, which either add latency or change phase response (== waveform shape)
Having full bandwidth without aliasing calls for a different method. Urs discussed it re: hive already
I like how zebra sounds; but you’re really barking up the wrong u-he tree imo.
Have you tried Hive?
-
- KVRist
- 186 posts since 30 Dec, 2008
Diva is great, but LIMITED. Modular synth like zebra with Diva sound engine - this is my blue dream. I need mod matrix with 100500 slots to modulate everything, at least 4 eg and 4 lfo, wavetables, phase modulation, 4 oscillators per layer, multiple layers, ect. ect. Unison for each oscillator section and global unison. And all this cpu friendly. Ahhhah...Ploki wrote: Sat Jul 18, 2020 12:29 am Uh...
diva already has all of that - including optional aliasing on the digital OSCs.
Motorolla DSP that Virus Ti uses - runs at 150 khz. We have now 10005000 core processors runs at 4 ghz, but current softsynth eats it all and want more... Why? I don't know, i'm not a dsp specialist...
Developers of the past prove that you don't need a supercomputer to make great sound...
Yes, and i don't like how Hive sounds in the HF domain, i don't know why. Just not that sound that i like... Closest word - lifeless? Too perfect? (Same for DUNE). Maybe i need some aliasing? Don't know...Have you tried Hive?
I tried everything, EVERY soft synth... They all sound the same basically in that aspect, i don't know why. The only synth that stand out of this is the Roland System-1 plugin. Don't like that synth overall (crappy interface, too limited in features and modulations, lack of stereo unison, hard digital sound overall), but i like how supersaw sounds on them! They have that bright "BZZZZZ".
-
- KVRAF
- 24415 posts since 7 Jan, 2009 from Croatia
Just get your 90s VA hardware and be done with it. You won't get that sound with modern softsynths, it's just not gonna happen.
What you need to understand is that by and large those old VAs used quite naive DSP algorithms for things. JP-8000 has a ton of aliasing, simply because it doesn't even do bandlimiting. You have this oscillator copied to its exactness in Diva.
ZebraHZ then. Diva filters inside.SoulState wrote: Sat Jul 18, 2020 12:56 amModular synth like zebra with Diva sound engine - this is my blue dream.
No, you don't. Just like how many, many, many Diva and Zebra patches by Howard Scarr show.SoulState wrote: Sat Jul 18, 2020 12:56 amI need mod matrix with 100500 slots to modulate everything
150 MHz not KHz. Also they are completely different CPUs with different instruction sets. Your computer's CPU is a general-purpose device that runs many more things simultaneously besides your softsynth. So, your OS and all the services it's running and all the background programs eat up into that CPU usage. It's just a different thing.SoulState wrote: Sat Jul 18, 2020 12:56 amMotorolla DSP that Virus Ti uses - runs at 150 khz. We have now 10005000 core processors runs at 4 ghz, but current softsynth eats it all and want more... Why? I don't know, i'm not a dsp specialist...
What you need to understand is that by and large those old VAs used quite naive DSP algorithms for things. JP-8000 has a ton of aliasing, simply because it doesn't even do bandlimiting. You have this oscillator copied to its exactness in Diva.
I think your ears are broken, quite plain and simple, if you call Hive's sound lifelessSoulState wrote: Sat Jul 18, 2020 12:56 amYes, and i don't like how Hive sounds in the HF domain, i don't know why. Just not that sound that i like... Closest word - lifeless? T
-
- KVRist
- 186 posts since 30 Dec, 2008
It is too expensive for me and lacks usability of virtual synths (many copies at once, etc). And also you need good converters to record it (which are not cheap too)...
Just don't like how it sounds. The problem not in filters i think, but oscillators...ZebraHZ then. Diva filters inside.
I better know what i need for my music))) I don't write soundtracks or that old analoguish synthwave and stuff. I write dance music with massive use of complex modulation and x1000500 unison on almost every sound.No, you don't. Just like how many, many, many Diva and Zebra patches by Howard Scarr show.
Hmmm my cpu usage when it idle is about 0 to 1 %.150 MHz not KHz. Also they are completely different CPUs with different instruction sets. Your computer's CPU is a general-purpose device that runs many more things simultaneously besides your softsynth. So, your OS and all the services it's running and all the background programs eat up into that CPU usage. It's just a different thing.
About what "background tasks" do you say?
Not exactness. Even with HQ turned off it is sounded DULL compared to the real one... For me it is a HUGE diffrence!What you need to understand is that by and large those old VAs used quite naive DSP algorithms for things. JP-8000 has a ton of aliasing, simply because it doesn't even do bandlimiting. You have this oscillator copied to its exactness in Diva.
The whole history of softsynths is trying to achieve that sound.
Withous success (well, almost). For 20 years.
If it is so simple and use naive algos, why here is no any authentic sounding softsynth, like jp8000, Nord, Virus? Such synth would instantly become a bestseller!
Glad for you! If you like that type of sound, just turn on any vst synth and be happy!I think your ears are broken, quite plain and simple, if you call Hive's sound lifeless
I think everyone ears is broken, including many software developers))
In my opinion there is great synths in sound aspect: Diva, Re-Pro series, Legend, Obsession, Ob-Extreme, Tyrell, Tal UNO Lx (in some tests sounds better than original btw;)), Diversion (but totally unusable due to very high cpu use), Cypher/Cypher2 and Strobes. That's all. Sylenth1 is good but dirty in hf, Spire is good but thin sounding overall, same for dune.
But here is one main problem - almost all that really good sounding soft synths has lack of features and not as versatile like spire etc. And HUGE cpu hogs!
Btw, Tyrell sound engine is good enought, maybe some filter and oscillator models include in Zebra3? (But it is a cpu hog also
)I think with some additional modulation features and unison that Zebra architecture delivers it can be my dream synth and a REAL competitor to that classic VA synths!
So this is my official feature request for zebra 3!
Btw, I read that Tyrell in first was a hardware VA Project, heh...
So i really biased towards that type of sound. Just add cpu optimisations, extended mod matrix, polyphonic stereo unison with detune knob and supersaws, and this thing became my dream synth...
Last edited by SoulState on Sat Jul 18, 2020 2:43 pm, edited 2 times in total.
-
- KVRAF
- 24415 posts since 7 Jan, 2009 from Croatia
No it's not. Not sure where you got that from. The very first commercial VST was trying to be an emulation of Minimoog, not a 90s VA. The first 3rd party VST was an emulation of Prophet-5, not a 90s VA. You got your marbles mixed up there.SoulState wrote: Sat Jul 18, 2020 1:57 pmThe whole history of softsynths is trying to achieve that sound.
Plenty of plugin vendors are quite successful without trying to chase "that" sound. Which should tell you something about "that" sound and what most people are looking for today in a softsynth. Let me give you a hint: most of them have moved on from that sound and embraced new sounds possible in softsynths that weren't possible on those old VAs.
There are many synths out there that didn't need to sound like any of the VAs you listed in order to become bestsellers. So you have a false dichotomy there.
Also... Dune sounding THIN?! Yep, your ears are definitely broken.
Oh and... if you're looking for Virus sound, there's Viper. Adam got it pretty darn close I'd say (and he's a part of that whole dance music scene).
Last edited by EvilDragon on Sat Jul 18, 2020 2:30 pm, edited 2 times in total.
-
- KVRist
- 186 posts since 30 Dec, 2008
So classic hardware VA synths is emulating analog synths also, what so?) There is no such difference to emulate analog or emulate virtual analog that emulates analog)))No it's not. Not sure where you got that from. The very first commercial VST was trying to be an emulation of Minimoog, not a 90s VA. The first 3rd party VST was an emulation of Prophet-5, not a 90s VA. You got your marbles mixed up there.
Do you related to dance scene somehow? (I think not really)
There is no problem for modern softsynths emulating minimoog or any other retro analog synth, there is some great emulations exist.
I talking about synths as dance workstantions, that is, basically same classic analog emulation engine with much more features (Massive unison with 128 voices per patch and cpu friendly at same time). So currently there is no one of that synth exist in the native VST format.
Imagine cloning Tyrell 128 times. That i talking about. You need a supercomputer to do that. That i call "no success".
Last edited by SoulState on Sat Jul 18, 2020 2:24 pm, edited 1 time in total.
-
- u-he
- 30206 posts since 8 Aug, 2002 from Berlin
Long after the factor presets were conceived, maybe some 10 or so years ago, we added the "crisp" render option to Zebra2's oscillators. It's been a remedy for all the young folks who perceived Zebra as lacking beyond 17kHz, at least with the Spectro/GeoMorph waveforms.
Unfortunately people seem to be more comfortable drawing Blend waveforms, which, with their 128 points may be easy to use, but they can't reach brilliance in bass notes. Unless you add some oscillator effects or plenty of distortion.
Zebra 3 will certainly have greater detail in its editable waveforms, but this comes with drawbacks in usability. Hence, some editors such as the built-in ones in some synths offer less control over detail and promote kind fo same-samey sounds. Consequently, we will offer some tools which go beyond any of it, much like we might have hinted at when we released Hive with its .uhm scripting language. Those results are outstanding.
-
- KVRAF
- 24415 posts since 7 Jan, 2009 from Croatia
Those two things don't go very well together. You can have either one or the other, but not both simultaneously. Calculating all those unisons WILL cost CPU the more features you add to the synth.SoulState wrote: Sat Jul 18, 2020 2:20 pmMassive unison with 128 voices per patch and cpu friendly at same time
Of course there is a difference, because those 90s VAs failed in emulating those vintage analogs correctly, whereas softsynths got much much better at it - so the end result is very different.SoulState wrote: Sat Jul 18, 2020 2:20 pmThere is no such difference to emulate analog or emulate virtual analog that emulates analog
-
- KVRist
- 186 posts since 30 Dec, 2008
But they somehow managed it to work with some ancient motorolla dsp's...Those two things don't go very well together. You can have either one or the other, but not both simultaneously. Calculating all those unisons WILL cost CPU the more features you add to the synth.
I say - emulations that deliver that old or not so old HW VA synths is enought for me and my tasks... Maybe it is not the exact or "correct" copy of some analog synths, but they sounding good anyway by themselves, translating some of that "analog character". But i can't say same for most softsynths, sadly. Often they just have no character at all... Too clinical and too, emm... "digital". Not inspiring.Of course there is a difference, because those 90s VAs failed in emulating those vintage analogs correctly, whereas softsynths got much much better at it - so the end result is very different.
-
- KVRAF
- 24415 posts since 7 Jan, 2009 from Croatia
Not a 128 voice unison, they didn't. Also they didn't even have massive unisons. Virus could only have 9x unison per voice. Sure you could stack voices but it would reduce your polyphony greatly. That's because they are working within known bounds of available CPU power that they have. A softsynth can always use a beefier CPU, which would make sure it can put out more polyphony.SoulState wrote: Sat Jul 18, 2020 2:57 pmBut they somehow managed it to work with some ancient motorolla dsp's...Those two things don't go very well together. You can have either one or the other, but not both simultaneously. Calculating all those unisons WILL cost CPU the more features you add to the synth.
BTW, as far as unison and CPU efficiency, it doesn't get better than Hive. It can do 64x unison per voice with least possible CPU spent, lower than any other softsynth.
-
- KVRist
- 186 posts since 30 Dec, 2008
Virus Ti have up to 16x unison per voice, not 9. 9 - is in the hypersaw oscillator. This is not "real" unison, just emulates that within osc.Not a 128 voice unison, they didn't. Also they didn't even have massive unisons. Virus could only have 9x unison per voice. Sure you could stack voices but it would reduce your polyphony greatly. That's because they are working within known bounds of available CPU power that they have. A softsynth can always use a beefier CPU, which would make sure it can put out more polyphony.
I talking about overall voices with polyphony, so 8 voices with 16x unison per voice = 128 total.
Camel audio Alchemy can much more in terms of unison, but sound just smearing instead of "fatting". In most softsynths it is like that... High harmonics became a pink noise... On the Virus (or An1x) it is not! You can clearly see that on the linear spectrogram. You can see that very high harmonics in the 15 khz range and over, that gives brilliance and presence to the sound, most vst synth produces just noise in this range...BTW, as far as unison and CPU efficiency, it doesn't get better than Hive. It can do 64x unison per voice at least possible CPU, lower than any other softsynth.
-
- KVRAF
- 24415 posts since 7 Jan, 2009 from Croatia
Yeah, of course. But a proper unison always costs polyphony in hardware (unless it's Nord Lead 3 which has full voice unison without stealing polyphony). In softsynths this does not need to be case (but it WILL cost you CPU).SoulState wrote: Sat Jul 18, 2020 3:26 pmVirus Ti have up to 16x unison per voice, not 9. 9 - is in the hypersaw oscillator. This is not "real" unison, just emulates that within osc.
I talking about overall voices with polyphony, so 8 voices with 16x unison per voice = 128 total.
Alchemy has up to 600 oscillators per voice, yep. By default, if you just change NOsc you will have all of them with the same phase. So you need to go to Edit tab and edit the phases first (just random dragging them around so they're not a straight line), then increase NOsc, and you get a much better sounding unison, that is not phasey. Takes a little bit of knowing how to use the synth before you start to shit all over it, y'know? Plenty other synths randomize phases on unison and this is not a problem at all.SoulState wrote: Sat Jul 18, 2020 3:26 pmCamel audio Alchemy can much more in terms of unison, but sound just smearing instead of "fatting". In most softsynths it is like that...
That is a ridiculous overgeneralization that is just not true.