8th European Conference on Speech Communication and Technology

Geneva, Switzerland
September 1-4, 2003


Microphone Array Voice Activity Detection and Noise Suppression Using Wideband Generalized Likelihood Ratio

Ilyas Potamitis (1), Eran Fishler (2)

(1) University of Patras, Greece
(2) Princeton University, USA

The subject of this work is the use of microphone arrays for speech activity detection and noise suppression in the case of a moving speaker. The approach is based on the generalized likelihood ratio test applied to the framework of far-field, wideband moving sources (W-GLRT). It is shown that under certain distributional assumptions the W-GLRT provides a unifying framework for evaluation of Direction of Arrival (DOA) measurements against spurious DOAs, probabilistic speech activity detection as well as noise suppression. As regards speech enhancement, we demonstrate the direct connection of W-GLRT with enhancement based on subspace methods. In addition, through the concept of directive a-priori SNR we demonstrate its indirect connection with Minimum Mean Square Error spectral (MMSE_SA) and log-spectral gain modification (MMSE_LSA). The efficiency of the approach is illustrated on a moving speaker where additive white Gaussian Noise (AWGN) is present in the acoustical field at very low SNRs.

Full Paper

Bibliographic reference.  Potamitis, Ilyas / Fishler, Eran (2003): "Microphone array voice activity detection and noise suppression using wideband generalized likelihood ratio", In EUROSPEECH-2003, 525-528.