I think I got interpolation between feedback coefficient matrices working (thanks to SciPy, which has polar decomposition, log and exp for complex matrices) but it blew up almost immediately, so I'm re-adding the normalization code in the hopes of getting some output that's more than a brief click followed by NaNs.
Trying to make morphing drones modulated by a third signal or something, not sure what it'll turn into...
Still not working properly. Brief tests shows the interpolation is probably fine (and the result doesn't blow up for $s = 0$ or $s = 1$), but as soon as $s$ is even slightly in between the filter becomes incredibly unstable and explodes, overflowing to infinity then NaN after about 0.2 seconds of output audio.
The interpolated matrices all have Frobenius norms smaller than the endpoints, but I don't know if that means anything.
It's also very slow, needing to do 5.5k matrix exponentials of 129x129 complex matrices for every second of output audio.
Abandoned interpolation/morphing. Doesn't work.
Made the current working part stereo.
claude