LOUD A 1020-NODE MICROPHONE ARRAY AND ACOUSTIC.pdf

ICSV14 Cairns ? Australia 9-12 July, 2007 LOUD: A 1020-NODE MICROPHONE ARRAY AND ACOUSTIC BEAMFORMER Eugene Weinstein 1 ? , Kenneth Steele 2 ? , Anant Agarwal 2 3 ? , James Glass 3 § 1 Courant Institute of Mathematical Sciences 251 Mercer Street, New York, NY 10012, USA 2 Tilera Corporation 1900 West Park Drive, Suite 290, Westborough, MA 01581, USA 3 MIT Computer Science and Artificial Intelligence Laboratory 32 Vassar Street, Cambridge, MA 02139, USA ?eugenew@ ?ken@ ?agarwal@ §glass@ Abstract Recording speech and other sound is difficult in environments with a large amount of noise and/or crosstalk. In these environments, array microphones are needed in order to obtain a clean recording of desired speech. In this work, we have designed, implemented, and tested LOUD, a 1020-node microphone array. To the best of our knowledge and as documented by Guin- ness World Records [6], this is currently the largest microphone array in the world. We have implemented an acoustic beamforming algorithm for sound source amplification in a noisy en- vironment, and have obtained preliminary results demonstrating the efficacy of the array. From one to 1020 microphones, we have shown a 13.7dB increase in peak SNR for a representative utterance, an 87.2% drop in word error rate (WER) with interferer present, and an 91.3% drop in WER without an interferer. 1. INTRODUCTION Speech recognition, and sound recording in general, in the presence of significant noise or crosstalk is difficult. When sound is recorded in a noisy environment through a single micro- phone, proximity of the microphone to the speaker’s mouth is essential for the high-quality recording needed for speech recognition. This proximity can not be achieved without tethered close-talking microphones. However, human friendly pervasive computing environments such as CMU’s Aura [4] or MIT’s Oxygen [14] are characterized by mobile users going about their daily business and preclude the use of tethered microph