語音應用於行動路徑導航之研究

Abstract

國內外發展汽車導航從最初期的筆記型電腦配置電子地圖光碟與導航軟體，發展至今已有成熟的車用導航產品，並且逐漸成為車輛標準配備。隨著地理資訊系統技術的演進，目前車輛導航系統可顯示詳盡的圖形資料於其配備之顯示器上，但圖形的複雜度直接關係到駕駛者將注意力分散到顯示器所引發的行車安全問題。另外，駕駛者在行車途中也可能引發操作導航軟體的需求，目前車用導航系統以按鍵操作軟體的方式，對於行車安全仍然造成影響。 本研究思考以語音作為汽車導航系統之人機介面，並實驗性地將語音辨識與語音提示功能加入其中，以達到提升行車安全的目的。在本研究中所使用的訊號處理技術，均是語音辨識領域中所慣用的，或者是計算量較少的，字詞範圍為特定詞辨識。從讀入錄音檔案開始，本研究將理論公式化為程式實作，達到使用者下達地圖操作指令後，軟體出現應對動作的目標，以及在轉彎路口前，以語音提示駕駛人詳細的轉彎資訊，包括路口距離、轉向角度、轉向後之路名。另外，亦具有設施提示功能，以語音告知駕駛者道路周邊之停車場、飯店、加油站等設施資訊。 將語音辨識應用於個人數位助理器上容易發生執行效能不佳的情況，尤其是語音辨識需要對語音訊號作一連串的處理，並比較各語音模型間，語音特徵值的失真量，使得辨識速度與硬體有直接關連。而本研究之語音提示使用文字轉語音技術，雖然能清楚達到轉向提示與設施提示兩大目的，但在遭遇破音字的情況下，有可能會產生不正確的發音，應是此系統未來應著重的問題。 關鍵詞：汽車導航系統、地理資訊系統、語音辨識、個人數位助理器

Vehicle navigation system has been developed from in-car notebook with digital map on CD-ROM to the advanced devices nowadays. At the same time, the progress of geographic information system enables the graphics on the screen to show details. However, the complexity of the digital map also draws driver’s attention as result of safety concerned. On the other hand, the manual operation of the navigation devices drives the same issue. The study takes the voice as a human-machine interface to improve the driving safety with speech recognition and voice response functions programmed on the personal digital assistant. The algorithms used in the study are referring to the method and theory in the key word speech recognition field. The navigation software built by this study has realized the speech recognition on the PDA to operate map controls, and the voice responses include the corner distance, angle, next route, the information of parking lot, hotel, and gas station. The major obstacle applying speech recognition is the efficiency of the program, a series process handling a voice file which slows down the speed of recognition. Although the users can get the voice responses clearly, the multiple pronunciations of a Chinese word may decrease the accuracy of it, and should be an issue in the future study. Key Words：Vehicle Navigation、Geographic Information System、Speech Recognition、Personal Digital Assistant