5aSP2. Small-scale matched filter array processing for spatially selective sound capture.

Session: Friday Morning, December 5


Author: Richard J. Renomeron
Location: Ctr. for Comput. Aids for Ind. Product., Rutgers Univ., P.O. Box 1390, Piscataway, NJ 08855-1390
Author: Daniel V. Rabinkin
Location: Ctr. for Comput. Aids for Ind. Product., Rutgers Univ., P.O. Box 1390, Piscataway, NJ 08855-1390
Author: Joseph C. French
Location: Ctr. for Comput. Aids for Ind. Product., Rutgers Univ., P.O. Box 1390, Piscataway, NJ 08855-1390
Author: James L. Flanagan
Location: Ctr. for Comput. Aids for Ind. Product., Rutgers Univ., P.O. Box 1390, Piscataway, NJ 08855-1390

Abstract:

Matched filter array (MFA) processing has a distinct advantage over delay-sum beamforming for reverberant enclosures in that it is able to cohere significant reflected images in addition to the direct arrivals and is capable of spatial selectivity in three dimensions. However, previous study of this technique utilized a very large number of sensors, which is beyond the A/D capacity and computational power of current commercially available DSP hardware. In this paper, the results of using matched filtering with only eight sensors in real rooms are presented. It is found that a moderate amount of reverberation can be removed, which suggests that the technique may prove useful for high-quality speech pickup for teleconferencing in a small-room environment.


ASA 134th Meeting - San Diego CA, December 1997