LOUD A 1020-node modular microphone array and beamformer for intelligent computing spaces.pdfVIP

LOUD: A 1020-Node Modular Microphone Array and Beamformer for Intelligent Computing Spaces Eugene Weinstein, Kenneth Steele, Anant Agarwal, and James Glass MIT Computer Science and Artificial Intelligence Laboratory 32 Vassar Street Cambridge, MA 02139 USA {ecoder,steele,agarwal,glass}@ Abstract. Ubiquitous computing environments are characterized by an unbounded amount of noise and crosstalk. In these environments, tradi- tional methods of sound capture are insufficient, and 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 novel 1020- node microphone array utilizing the Raw tile parallel processor architec- ture [1] for computation. To the best of our knowledge, this is currently the largest microphone array in the world. We have explored the uses of the array within ubiquitous computing scenarios by implementing an acoustic beamforming algorithm for sound source amplification in a noisy environment, 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 on a representative utterance, an 87.2% drop in word error rate with interferer present, and an 89.6% drop in WER without an interferer. 1 Introduction The interaction between humans and computers has been a central focus of ubiq- uitous computing research in recent times. In particular, communication through speech has been extensively explored as a method for making human-computer interaction more natural. However, computer recognition of human speech per- forms when a recording can be made without the presence of much ambient noise or crosstalk. Seeking to create a natural setting, ubiquitous computing environ- ments tend to fall in this category of situations where natural-interaction speech recognition is a challenging problem. When significant levels of noise are present, or several humans are talking at the same t