| 摘要 |
<p>PURPOSE:To save the trouble for a user to perform the dictionary switching work by providing a recognition object control means which takes only voice pattern dictionary data corresponding to character string data used in an operating system as the recognition object at the time of starting the system. CONSTITUTION:A recognition object control means extracts the feature quantity of voice data inputted from a voice input part 10 by a feature quantity extracting part to generate a feature pattern, namely, voice pattern dictionary data (a) and character string data (b). Voice pattern dictionary pattern (a) is stored in a voice pattern dictionary data storage part 14, and character string data (b) is stored in a character string data storage part 15. When switching of voice dictionary data (a) and character string data (b) is necessary at the time of starting the operating system, this state is detected to automatically read out dictionary data corresponding to this state, and data is switched. Thus, the user is saved the trouble to perform the dictionary switching work.</p> |
