Outbreaks of forest pests cause large-scale damages, which lead to significant impact on the ecosystem as well as the forestry industry. Current methods of monitoring pest outbreaks involve field, aerial and remote sensing surveys. These methods only provide partial spatial coverage and can detect outbreaks only after they have substantially progressed across wide geographic areas. This paper presents an IoT system for real-time insect infestation detection using bioacoustic recognition via machine learning techniques. Specifically, we focus on detecting the Mountain Pine Beetle (MPB), which is the most destructive insect of mature pines in western North American forests. We present the design of the system and describe its various hardware and software components. Experimental results collected from a prototype implementation of the system are presented, which show that the system can detect MPB with 82% accuracy. We also demonstrate the applicability of our system in other noise monitoring applications, and report our experimental results on urban noise detection and classification.