Modulating formant filters
  • Attached is my first effort in Audulus. It currently plays a saw tooth wave through two formant filters (the "Ah" and "Ee" formants), with cross fade control set by an envelope filter. The trigger in latch mode turns on/off the sawtooth wave, while the other trigger launches the ADSR which modulates the cross fade.

    Each formant uses three parallel band pass filter chains to isolate a particular fundamental frequency. Each chain has three identical band pass filters in series to accentuate the effect.

    The "Ah" sound seems much nicer than the "Ee" sound so I'll have to check the parameters.

    I would welcome any feedback and guidance on next steps. I plan to add other modules for a couple more vowel sounds.

    But ultimately I would like this to be an effect applied to a guitar.

    (1) How do I pick out the start of a newly played note / chord to trigger the ADSR? Right now I just use a trigger output.

    (2) If I had four vowels set up and wanted to switch between vowel pairs (e.g. select between "ah" to "ee" vs. "uh" to "eh", how would I do that? In a traditional programming language, I would set up a parameter array for each option, with a 0 or 1 value indicating whether a particular formant was on in the "a" or "b" position of the cross fader. But I am less sure how to do arrays in Audulus.

    Darby.Formant v3.audulus
  • This is great! The Ah formant sounds like an over-the-top opera singer bahaha so good.

    As for your first question, I'm uploading a guitar synth tutorial today or tomorrow that will show you how to do all of that and more.

    Switching between two pairs would just be adding a crossfade node here (see attached patch - also, cleared the history and it's a little smaller now).
    Darby.Formant v3 - Bimini Road Edit.audulus
  • how do i find attachments to listen or watch
  • @btayloragent - are you on mobile? if so, look for the "full site" button at the bottom of the page - click on that and you should be able to see the attachments now.
  • I made a version with a few optimizations and enhancements, i took out the copied biquad bandPass filters and instead only copied the biquad object inside them so that the complex CPU heavy coefficient calculations are done only 1 time per filter since they are all the same. I also realized fading the audio between 2 copies did not actually create a realistic effect as when you move your mouth the formats are shifting between each other, and to realize that I made another optimization of having only the knob settings for different formats packed into poly objects and then put through a few mapped cross faders and then out to the actual filter.
    Darby.Formant_v5 MacroMachines Edit.audulus
    2732 x 2048 - 362K
  • This is really cool! Love the unison module there too that's nice
  • This last patch one of my favs. Thanks, it's very nice how things are done and about 10% on ipad air 2. Love the sounds!