Open sowbug opened 2 years ago
I've done some tuning of this parameter for the 12db LPF. It's still mysterious to me, but for most cases it sounds OK. Removing from current milestone.
This issue is now a placeholder for getting the linear/log story straight. See Pirkle DSSPC++ p.71 for MMA's simplified formulas for concave/convex transforms (the ones with the 5/12 fraction).
Pirkle recommends a single user-facing range of 1 to 10 for Q. DSSPC++, page 173.
I've read RBJ's Audio EQ Cookbook, and unfortunately "the EE kind of definition" means nothing to me. I can see that for the LPF given in that document, a Q of 1/sqrt(2), or 0.707, means no resonance, with the spectrum starting to drop just a bit below the cutoff frequency and continuing to drop at the expected 12dB/octave. A higher Q produces resonance (a peak around the cutoff). But I can't tell other than empirically where a higher Q stops delivering bang for the buck.