Yes, it would require some math, but I suspect you'd get superior results. For example, you can replace the "modeling" by the FFT transform applied to small time-slices (i.e., this determines which hair-cells get triggered at a given time). Now you have to stitch these slices together without introducing spurious frequencies, which is the difficult part.