Summary of time-frequency algorithms for processing auditory evoked potentials

Adaptive filters are extremely useful for signal analysis. Using a time-frequency viewpoint, adaptive filters can be designed for the short-term spectral analysis of signals. In this study, time-varying adaptive filters developed from the short-term Fourier transform are used for estimation of auditory evoked potentials (AEP). Six methods using different noise definitions were investigated for time-varying class 1 Wiener filter design and their performance was compared. Conventional time-invariant one-dimensional Wiener estimation filters were also designed for comparison. It was found that the filter using the difference of two subgroups´ average of post-stimulus recordings as noise gave the best result. Most time-varying Wiener filters perform better than the time-invariant ones designed using the same method.