Yeah... I like them as is... plus they are stereo and pretty flexible. They have a huge pitch range!
Bitwig The Grid: It is just me...
-
- KVRist
- 58 posts since 8 Jun, 2006 from Ann Arbor, MI
Agreed. Having come from the other side where all the oscillators are rich and warm, I much prefer being able to start clean and fatten things up. All you have to do is find a pleasing combo for thickening things up and save a preset to start from the next time.
-
- KVRAF
- 8850 posts since 6 Jan, 2017 from Outer Space
I haven't found better oscillators in any synth. But if you don't modulate the shape its as boring (plastic) as any other synth with basic waveforms. In the Grid, its easy to add some drift btw... (the one thing it needs to make an oscillator behave more analogic...)zerocrossing wrote: ↑Wed Sep 30, 2020 4:58 pmThis is why I spent a day kicking its tires and then never used it again.Echoes in the Attic wrote: ↑Sun Nov 03, 2019 1:19 pm Yep, amazing design. Just wish the oscillators didn't sound like plastic toy oscillators. A more analog sound like Reaktor blocks oscs would be cool.
Regarding analog waveforms, if you shape the sine, you do get an analog like sawtooth - compare the waveforms... Still not enough? Send it into some waveshaper/distortion...
And such an oscillator mod should be put into a container! That's why we need it!!!
-
- KVRAF
- 25635 posts since 3 Feb, 2005 from in the wilds
Yeah, they start pitch perfect... including left/right... a useful starting place for doing audio rate modulation. Then bend them from there and as you say, save a few preset templates as starting points.dehuszar wrote: ↑Thu Oct 01, 2020 5:37 pm Agreed. Having come from the other side where all the oscillators are rich and warm, I much prefer being able to start clean and fatten things up. All you have to do is find a pleasing combo for thickening things up and save a preset to start from the next time.
They can sound quite good and complex. And a bit of saturation by driving a filter adds some dynamic character that one expects from a classic osc, filter, vca setup.
-
- KVRian
- 1262 posts since 15 May, 2002 from Finland
I think it would have been very bad design to make the oscillators coloured in a general - usage modular. The shape does not affect CPU since AFAIK most synths use lookup tables anyway.
I would welcome the option for already coloured oscillators though, even if it's true that you can modulate them to be more interesting!
I would welcome the option for already coloured oscillators though, even if it's true that you can modulate them to be more interesting!
-
- Banned
- 11467 posts since 4 Jan, 2017 from Warsaw, Poland
This always fascinated me. Why not simply calculate the waveform on the fly? For basic shapes it's a simple math and you avoid all the issues with aliasing, etc. I wonder what Bitwig's doing actually.
-
tor.helge.skei tor.helge.skei https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=152647 - KVRian
- 528 posts since 30 May, 2007
-
- KVRist
- 216 posts since 31 Oct, 2015
Simple math means perfect waveforms which means (except for sine, but this one is not simple math) infinite numbers of frequencies and then aliasing.
-
- Banned
- 11467 posts since 4 Jan, 2017 from Warsaw, Poland
What are you talking about? Every programming language will have libraries with popular math functions, sine being one of them. It's trivial. Also I don't know what you think aliasing it, but precisely having "infinite" numbers to represent something is the opposite of aliasing. Aliasing happens when you try to represent high frequency content with a low frequency source.SebAV wrote: ↑Sun Oct 04, 2020 7:35 amSimple math means perfect waveforms which means (except for sine, but this one is not simple math) infinite numbers of frequencies and then aliasing.
-
- KVRian
- 513 posts since 26 Nov, 2009
Dude, stop talking, please. You want efficient realtime synthesis, so you can actually use your computer for music production, not for just for sound generation... Don't worry, you are not smarter than current generation of software developers.
https://en.wikipedia.org/wiki/Sine#Soft ... mentations
-
- Banned
- 11467 posts since 4 Jan, 2017 from Warsaw, Poland
"Dude". I never said I was smarter than anyone else. I'm as stupid as they come. But it doesn't change the fact you can easily calculate sine nowadays instead of reading it from a look-up table. Also not sure if you're aware, but EVERY sound can be replicated by a combination of sines of different frequencies, amplitudes and phasesanomandaris1 wrote: ↑Sun Oct 04, 2020 9:18 amDude, stop talking, please. You want efficient realtime synthesis, so you can actually use your computer for music production, not for just for sound generation... Don't worry, you are not smarter than current generation of software developers.
https://en.wikipedia.org/wiki/Sine#Soft ... mentations
-
- KVRian
- 513 posts since 26 Nov, 2009
In what way it's easy to compute? - it's a transcendental function, so (quoting from wikipedia):"A transcendental function is an analytic function that does not satisfy a polynomial equation, in contrast to an algebraic function. In other words, a transcendental function "transcends" algebra in that it cannot be expressed in terms of a finite sequence of the algebraic operations of addition, subtraction, multiplication, division, raising to a power, and root extraction."
I am also not a programmer or an expert in mathematical analysis, but I am sure that modern software is probably written as efficiently as it can get!
-
- KVRian
- 1262 posts since 15 May, 2002 from Finland
Yes, you should read the link already provided about and think about it a bit. You can also read this and the follow-up post on band limiting :
http://www.martin-finke.de/blog/article ... waveforms/
The fundamental problem is that reality can't represent mathematics. The perfect waveforms you are probably thinking about on can only exist at certain integer ratios.
Take a piece of grid paper, and try to draw a symmetric saw wave at the period of five x pixels for example. You can't.
I read recently a paper that you'd need 64x oversampling and a subsequent filtering (remember, this would be needed note) to get rid of perceptible aliasing with simple waves.
A common solution is to pretreat the waves so that no frequencies above the nyqyist frequency are present and put them in a lookup table. Then you just need the phase amount and you can pick up the sample you need. It's hard to imagine making the process more efficient really.
The notion about additive synthesis you made is correct, but to do it in realtime, you'd need to do hundreds of sine wave functions for each sample. That's millions of sine functions per second, and that's not what you want.
http://www.martin-finke.de/blog/article ... waveforms/
The fundamental problem is that reality can't represent mathematics. The perfect waveforms you are probably thinking about on can only exist at certain integer ratios.
Take a piece of grid paper, and try to draw a symmetric saw wave at the period of five x pixels for example. You can't.
I read recently a paper that you'd need 64x oversampling and a subsequent filtering (remember, this would be needed note) to get rid of perceptible aliasing with simple waves.
A common solution is to pretreat the waves so that no frequencies above the nyqyist frequency are present and put them in a lookup table. Then you just need the phase amount and you can pick up the sample you need. It's hard to imagine making the process more efficient really.
The notion about additive synthesis you made is correct, but to do it in realtime, you'd need to do hundreds of sine wave functions for each sample. That's millions of sine functions per second, and that's not what you want.
-
- KVRian
- 1262 posts since 15 May, 2002 from Finland
From : https://www.google.com/url?sa=t&source= ... OlAeiv9eKQ
We find an image that shows how much aliasing trivial (non treated) basic waveforms have :
https://ibb.co/mqG6gBD
This is another, more easy to understand article on the problem: https://www.vast-dynamics.com/?q=node/181
It's important to realize that in the physical world, as far as we know now, everything including time is quantised. So even the seemingly smooth waves from an analog oscillator can't escape this inaccuracy in comparison the the idea of a perfect waveform that can only exist in the mathematical reality.
We find an image that shows how much aliasing trivial (non treated) basic waveforms have :
https://ibb.co/mqG6gBD
This is another, more easy to understand article on the problem: https://www.vast-dynamics.com/?q=node/181
It's important to realize that in the physical world, as far as we know now, everything including time is quantised. So even the seemingly smooth waves from an analog oscillator can't escape this inaccuracy in comparison the the idea of a perfect waveform that can only exist in the mathematical reality.
-
- Banned
- 11467 posts since 4 Jan, 2017 from Warsaw, Poland
anomandaris1 wrote: ↑Sun Oct 04, 2020 11:36 amIn what way it's easy to compute? - it's a transcendental function...
Jeez, guys. I'm not suggesting VST devs should calculate the ideal physical / analog sinewave (or whatever other shape) and open a wormhole in the process!Taika-Kim wrote: ↑Mon Oct 05, 2020 6:49 amIt's important to realize that in the physical world, as far as we know now, everything including time is quantised. So even the seemingly smooth waves from an analog oscillator can't escape this inaccuracy in comparison the the idea of a perfect waveform that can only exist in the mathematical reality.
I'm obviously saying to quantise the results of such an ideal function to the required and sufficient "resolution" dictated by audio rate vs. bit depth grid. Alternatively they could pre-calculate number of waveform tables for many combinations of pitch at given audio rate / bit depth when the VST is 1st run and then read & interpolate from that. Perhaps that's what @Taika-kim meant in the original post I mentioned?
BTW, for example ArcSyn claims to be calculating the oscillators in realtime:
https://www.spcplugins.com/arcsyn-manual-osc-waves
