SYSTEM FOR GRASPING SPEECH MEANING OF RECORDING AUDIO DATA BASED ON KEYWORD SPOTTING, AND INDEXING METHOD AND METHOD THEREOF USING THE SYSTEM

摘要

A key word extraction based speech recognition system for recorded voice data and an indexing method and a speech recognition method using this system are disclosed. The indexing unit of the system receives speech data and performs phoneme-based speech recognition on a frame-by-frame basis to form a phoneme lattice. The indexing information for the time-limited frame composed of a plurality of frames, Including phoneme grids formed for each frame of the time limit, and stores them in the indexing database. The search unit searches a phoneme string corresponding to the search word through a phoneme-by-phoneme comparison of the divided indexing information stored in the indexing database, using the key word input from the user as a search word, and performs a precise acoustic analysis on the matched phoneme string And the grasping unit grasps the main word through the search result searched by the search unit and outputs the grasped speech data to the user so as to grasp the spoken contents of the voice data.