Instant Learning Sound Sensor: Flexible Real-World Event Recognition System for Ubiquitous Computing

We propose a smart sound sensor for building context-aware systems that instantly learn and detect events from various kinds of everyday sounds and environmental noise by using small and low-cost device. The proposed system automatically analyzes and selects an appropriate sound recognition process, using sample sounds and a parameter templates database in the event learning phase. A user is only required to input target event sounds from a microphone or sound files. Using the proposed sensor, the developer of ubiquitous service can easily utilize real world sounds as event triggers to control appliances or human’s activity monitors for presence services without a signal processing programming.

• Type: Conference
• Year: 2007
• Where: UCS
• Authors: Yuya Negishi, Nobuo Kawaguchi
• Keywords: Applied Computing, Appropriate Sound Recognition, Everyday Sounds, Smart Sound Sensor, UCS 2007