摘要 |
PURPOSE: A voice dividing and recognizing device, and a method thereof are provided to perform voice recognition by properly dividing inputted voice according to a speaking speed and purpose of a speaker. CONSTITUTION: A voice divider(60) outputs first divided voice data by dividing a voice frame data row outputted from a voice processor(40). If the first divided voice is longer than the length corresponding to a speaking speed and a voice recognition purpose, the minimum window frame energy value is searched. The front end of a frame having the minimum window frame energy value is appointed for the end point. The end of the frame is appointed for the start point. The voice divider outputs second divided voice data by performing secondary division based on the end/start point. A voice recognizer(70) performs voice recognition through the first/second divided voice data.
|