Intelligent appliance control using a low-cost embedded speech recognizer

摘要

In this paper, we present an embedded system design for speech recognition based appliance control. The Mel-frequency cepstral coefficient (MFCC) features for speech attribute extraction and dynamic time warping algorithm (DTW) for speech attribute matching are implemented with the low cost and resource-limited micro controller unit (MCU) - SPCE061A. While sampling input speech data, the idle time of MCU is used to perform MFCC extraction and DTW for frame-synchronous operations to achieve real-time requirement. The proposed embedded system can be used for the speech based appliance control which is a key technology for home automation.