Digital infinite impulse response (IIR) filters are popular in audio signal processing applications because they typically have a much lower computational cost than finite impulse response (FIR) filters designed to the same specifications. However, IIR filter design is difficult, particularly for arbitrary frequency responses that span multiple octaves, as is common in audio. A recent technique known as frequency warping offers a new way to design arbitrary IIR filters with much greater accuracy than before.
Dr. Kite will review existing IIR design methods, and show how frequency warping is related to the popular bilinear transform. Using a multi-band approach pioneered by Bank, he will show how IIR filters can be designed for the entire audio band with accuracy limited only by filter complexity. Dr. Kite will demonstrate a 30-pole filter designer controlled via a GUI, and show how high-order filters can alleviate common audio headaches such as the ragged frequency responses of loudspeakers.