基於語音辨識之智慧型輪椅控制

本研究將利用Raspberry Pi嵌入式平台開發一套語音辨識系統，搭配外部的立體聲麥克風以及外部音效裝置構成系統之硬體架構應用於電動輪椅上。首先，利用語音辨識演算法，將聲音訊號進行訊號前處理再求取訊號之特徵參數。除此之外，利用Fuzzy c-means模糊C平均值演算法進行語音分群比對，使其能準確辨識出特定語者之語意來決定輸出之電壓以控制電動輪椅。最後經由實驗與結果分析，實際驗證本語音系統辨識率的精確性及可靠性。

A voice recognition system employing Raspberry Pi embedded platform for intelligent wheelchair control is developed. By utilizing the voice pre-processing algorithm, the characteristic parameters of voice signals can be obtained. In addition, using the fuzzy c-means clustering algorithm, the voice signals can be classified to recognize the specific language semantics accurately and then to provide the input voltage for robust control of the intelligent wheelchair. To verify its feasibility, an experiment was performed to prove the control accuracy and reliability of the developed voice identification system.