View Single Post
Old 01-31-2017, 02:54 PM   #34
earlevel
Human being with feelings
 
Join Date: Dec 2015
Posts: 331
Default

The solution is pretty much as you guessed, in general—smooth the parameters by processing their filters and calculating fresh coefficients. If that's a problem, there are things you can do to lower the processing required. You can lower the "control rate" by updating the smoothing filters and calculations every 10 samples, for instance. Or, use a parameterized filter—where the frequency, Q, and gain don't interact (state variable filter or Moog-style filter made of one-poles and feedback). First, that means that you don't need to recalculate Q when changing frequency only. Also, it makes it more practical to smooth the coefficients instead of smoothing the parameters and calculating coefficients every time. (That is, when the frequency knob moves, calculate the new frequency coefficient and run that through a smoother that's processed in the audio routine—then use the smoothed coefficinet output directly, avoiding the calculation at audio rate.)
earlevel is offline   Reply With Quote