Author |
Message |
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Wed Nov 23, 2016 5:09 pm Post subject:
Trivial oscillators Subject description: Forcing stock G2 oscillators to generate trivial non-bandlimited waveforms |
|
|
Hi,
Long time ago I made the (baffling) discovery that signals hitting the FM inputs of stock G2 oscillators were ignored by their internal band-limiting (anti-aliasing) processes.
What I discovered just lately is that, by dialing an oscillator into 0Hz mode, using FM Track mode and an accurately scaled fixed FM offset, one thus can force a stock G2 oscillator module to generate its waveform naively, meaning without any bandlimiting (anti-aliasing).
Of course this isn't at all useful when utilizing bespoke oscillator as a primary audio source. However, when using it to drive a nonlinearity in order to obtain new waveshapes, the non-bandlimited version performs much cleaner and better, as it doesn't contain the added BLEP/BLAMP residuals which generate unwanted artifacts when processed through the nonlinear waveshaper.
The attached building block patch contains a comparison scheme, which clearly demonstrates the audible advantage of a non-bandlimited G2 sawtooth oscillator over a bandlimited one when driving a control-sequencer used as a fiercely nonlinear waveshaper.
This discovery also renders obsolete my building block deriving a non-bandlimited audio-rate sawtooth from a sawtooth LFO via audio-rate inter-/extrapolation, because it's obviously way simpler and cheaper to use a stock sawtooth oscillator set up as demonstrated in the patch here. Hence quite a few of my previously published patches, which use sawtooth LFOs in the aforementioned manner, could be improved accordingly. Ah, "there is no end to learning". (-Robert Schumann)
cheers,
t
Description: |
Forcing stock G2 oscillators to generate trivial non-bandlimited waveforms |
|
Download |
Filename: |
TrivialOsc____TK.pch2 |
Filesize: |
1.79 KB |
Downloaded: |
3745 Time(s) |
|
|
Back to top
|
|
|
jamos
Joined: Jun 01, 2004 Posts: 514 Location: Eugene, Oregon, USA
Audio files: 4
G2 patch files: 41
|
Posted: Sat Nov 25, 2017 12:51 pm Post subject:
|
|
|
Tim, I have to admit... I can't hear any significant difference between the two versions. Both have substantial aliasing artifacts. Is there an error in the patch?? |
|
Back to top
|
|
|
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Tue Nov 28, 2017 5:00 am Post subject:
|
|
|
Hi Jamos,
I don't own a G2 anymore, but the patch performs as intended on the demo version.
Both versions will exhibit aliasing since the whole process of using an unipolar sawtooth oscillator to read out of a control sequencer isn't bandlimited. However, forcing the oscillator to generate its sawtooth naively as demonstrated in the patch will produce a cleaner result.
Using a control sequencer as a waveshaper probably isn't the best example because of the lack of bandlimiting as mentioned. However if you use a 0Hz PhaseMod sine oscillator as a sine waveshaper and drive it with the correct mod range, the difference becomes night and day. Using a "normal" sawtooth (bipolar this time) as the modulator will exhibit artifacts generated by the bandlimited flanks. The trivial one will generate a pure and perfect sine. |
|
Back to top
|
|
|
jamos
Joined: Jun 01, 2004 Posts: 514 Location: Eugene, Oregon, USA
Audio files: 4
G2 patch files: 41
|
Posted: Wed Nov 29, 2017 6:26 pm Post subject:
|
|
|
OK Tim, I'll give that technique a try. Thanks. |
|
Back to top
|
|
|
|