Hi. As of now, we dont have to make oscillators that can switch between waveforms (i could combine 2 or 3 oscillators with a switch, but that will increase CPU). Is wavetables the way to go instead?
I have samples from a Roland Jupiter 4 that i could use for different waves, but what about "stretching" the waveform up and down when i play a high/low key. Wont i get bad sample stretching/compression?
Thanks:)
Essentially the weak property of wavetables is that if they're played faster than their original pitch, harmonics that move above the nyquist frequency (half of the sample rate, 24kHz) will appear at a different frequency - aliases. And when playing slower than their original pitch, harmonic content will be missing, at quarter speed, there will be no harmonics above 6kHz.
Both are annoying when creating harmonically rich waves like square, saw, triangle, pulse waves.
There are some algorithms that mitigate the first problem, like "pre-integrated wavetables", but that's a developer's topic rather than a patching technique.
Yeah.... I wondered about aliasing etc. Thanks for explaining it. I will just wait til a multiwave osc is released instead.
Thanks;)
You can derive a square/pulse wave from a saw or triangle with the dist/inf object, but I don't know how well that would hold up against the dedicated oscillators in terms of aliasing and things. You get the pulse waves by feeding the base waveform into the dist/inf along with an offset.
"dist/inf" estimates the zero-crossing location by linear interpolation between two successive incoming samples, and then uses a 64x oversampled edge for the output - the same as "osc/square" and "osc/pwm".
"osc/square" and "osc/pwm" know the exact locations of the edges, rather than estimating.
At high frequencies the linear interpolation assumption becomes inaccurate, but it does a lot better than direct clipping in the time domain.
My favorite use is
oscillator->resonant filter->(offset)->"dist/inf"
like demos/synth/screamer
Ah yeah, I can sort of imagine what that would sound like. With high resonance you'd get a sort of multipulse waveform... kind of sync-y sounding I guess?