A Speech Indicator for the VoWSN approach

In this paper we define and evaluate a Speech Indicator (SI) basing on acoustic data extracted by a Wireless Sensor Network (WSN) in a noisy environment. This way we want to exploit the potential benefits deriving from Voice over Wireless Sensor Networks (VoWSNs). A WSN topology for audio surveillance is proposed in order to identify and quantify voice presence and to activate a noise cancellation algorithm in case of scarce intelligibility. Moreover, a proper tradeoff between audio processing issues and WSNs capabilities is discussed.