Adaptive Mehrkanalgeräuschunterdrueckung und Sprecherortung bei gestörten Sprachsignalen innerhalb geschlossener Räume

Authors: P. Kuczynski, S. Fischer, F.-C. Kruegel, A. Wasiljeff, K.U. Simmer
Abstract: We present a noise reduction system for suppression of incoherent noise sources and reverberation of disturbed speech signals within closed rooms. The system is based on a beamformer with a Wiener filter at the output and an adaptive beam steering unit that can follow the speaker's movements. This system presented here uses two Wiener filters. The first one is combined with the time delay estimation using the generalized cross correlation function. The second one is based on a algorithm presented by Zelinski. Simulations of the room impulse response show that the time delay estimation becomes much more precise in presence of reverberation by using the generalized cross correlation function. Further simulations prove the better noise reduction of this algorithm over competitive systems. In order to verify the simulations the noise reduction system was implemented in real time on a signal processor board using four input channels. The noise reduction and speaker localization capabilities are good even in realistic situations with one or more noise sources and disturbance by other persons ("Cocktail Party Effect") if the speaker is close to the microphone array.
Document type: Conference Paper
Publication: 9. September 1995
Conference: Kleinheubacher Berichte
Pages: 369-378
Volume: 38
Index: 79
Files: BibTEX
Last change on 29.04.2008 by Admin
AIT ieee GOC tzi ith Fachbereich 1
© Department of Communications Engineering - University of BremenImprint / Contact