Efficient pole-zero modeling of resonant systems using complex warping and Kautz filter techniques

The paper describes new methods for designing pole-zero filters that utilize a particular filter structure called Kautz filters and an efficient procedure for the optimization of filter parameters. In the design phase, complex warping is used to emphasize a chosen frequency region in the modeling. The motivation, from an audio signal processing perspective, is as follows. Many audio related target responses can be well-modelled by a combination of distributed decaying exponential components, which is by definition what a Kautz filter does, but in an orthonormalized form, providing many favorable properties. The most essential part is then how to choose the Kautz filter poles, for which we propose a new effective and well-behaving iterative method. Utilizing an intermediate warping procedure, the desired frequency resolution allocation is incorporated into the choosing of poles $the target signal is warped prior to the pole optimization and then the produced poles are mapped back to the original frequency domain using inverse warping.