家用機器人之語音辨識系統

Abstract

本論文發展出一套可應用於家用機器人操控之語音辨識系統。為建立方便之人機介面，文中提出一針對家用機器人應用之語音辨識系統的架構，並運用德州儀器公司生產之TMS320VC5402 DSK Board作為運算平台，建立基於動態時間扭曲(Dynamic Time Warping)演算法之語音辨識系統。本論文已完成一平均辨識成功率可達90%的系統，並成功地整合於家用機器人之上，可針對家用機器人作即時性的語音操控動作，成為一有功效之人機介面。

In this thesis, we developed a speech recognition system for a home robot. We propose a home robot human-machine interface architecture based on the speech recognition system. The speech recognition system is implemented on a TI TMS320VC5402 DSK board. We take advantage of DSP chip and build our system based on the Dynamic time warping (DTW) method. We successfully achieved a 90% average recognition rate. This system has been integrated with a self-made home robot. Experimental results show that the speech recognition system can become an effective human-machine interface for a home robot.