User Manual

37 User Manual v 1.0.0
UNFILTER
Understanding LEARN
Overview
The LEARN function lies at the heart of UNFILTER, and thus understanding it will help you
getting the most out of the plug-in.
Well, how does it work?
LEARN first creates a distinction between spurious frequencies in your signal (such as noise),
and long term stationary components that contribute more to the overall sound of the
recording. Once this distinction has been made, the stationary components are evaluated over
time to see which components change rather quickly (= most likely part of an instrument or
voice) and which components do not.
For instance, filtering effects such as room resonances tend to stay stationary or change only
slowly with time when compared to the actual frequencies in a musical performance. This
distinction in time and frequency allows the UNFILTER algorithm to separate out a frequency
response curve that can estimate and successfully remove an undesired filtering effect.
Since both the frequency response of the undesired filtering and the actual filtered signal are
unknown to the algorithm in advance this process is called "blind deconvolution".