Trapezoid Oscillator with variable slope
  • I have been wishing for a variable slope trapezoid wave for a while and I finally thought of a way to use the fract expression and the crossfade node to make a symmetrical waveshaping function. I am hoping to use this a building block for some other fun wave-shaping oscillators.
    image
    Screen Shot 2017-09-02 at 12.23.11 PM.png
    957 x 515 - 73K
    trapezoid osc.audulus
    19K
  • Neat approach! Very clean way to morph from a triangle to a square wave without changing frequency, phase or amplitude. The phasor is a great starting point for all kinds of oscillators. I would love to see Taylor introduce an anti-aliasing node that could be inserted just after the oscillator module. That way you would have the mathematically simple phasor ramp to play with, without introducing any aliasing artifacts.
  • Oh an anti-aliasing node would be nice, although for my needs the current set-up still looks pretty good on an oscilloscope and sweet to my ears. Also something to consider would be that there would still be some distortion of the waveforms at the higher frequencies as such a filter will typically either permit some aliasing to occur or else attenuate some in-band frequencies close to the Nyquist limit (wikipedia). Maybe it will be less important as sample rates improve and we can internally oversample.

    bonus: quad trapezoid oscillator.
    Quad Trapezoid.audulus
    49K
  • The waveform distortion is why I thought it would be better to have it as a separate node rather than built into the phasor. If the phasor itself were anti-aliased then the ramp would necessarily be distorted at the transition points. If it were a separate node you could choose where in the signal path to apply it.
  • Taylor and I are talking about supersampling in Audulus...it's just a thought at the moment but we might have a mode where you can run Audulus at a super duper DUPER high sample rate and create (nearly) alias-free oscillators. So you wouldn't need an anti-alias node (and you actually wouldn't need z-1 either!). I'll keep you update about that - Taylor's doing it in what could be some proprietary way so I don't want to overreach. But anyway it's on our radars :)

    That said, to me, I feel like most of the alias sound as-is is either live-with-able or sounds good going into a filter. So at least it's not terrible at the moment! :)