Institut für Angewandte Physik, TH Darmstadt
Schloßgartenstr. 7, 64289 Darmstadt, Germany
Noise reduction is a problem being encountered in a variety of applications, such as environmental noise cancelation, signal recovery and separation. Passive noise reduction is done with the help of absorbers. Active noise reduction includes the transmission of phase inverted signals for the cancelation. This paper is about a threefold active approach to noise reduction. It includes the separation of a combined source, which consists of both a noise and a signal part. With the help of interaction with the source by scanning it and recording its response, modeling as a nonlinear dynamical system is achieved. The analysis includes phase space analysis and global radial basis functions as tools for the prediction used in a subsequent cancelation procedure. Examples are given which include noise reduction of speech.