摘要 |
PURPOSE: A selective speech recognition device in accordance with network situations and a method thereof are provided to perform a speech recognition function by using a self speech recognition engine and a remote speech recognition server, thereby using minimum resources only while realizing a fast recognition time and accuracy. CONSTITUTION: A user interface receives a user command to perform the user command, and informs a user of performed results. A speech feature extractor(30) extracts features of an uttered speech during a speech recognition. A speech recognition engine(40) carries out a speech recognition function in accordance with the speech features in a local speech recognition mode, and informs the user interface of carried results. A switching unit transmits the speech features to a remote speech recognition server in a distributed speech recognition mode when a network state is good, informs the user interface of recognized results, and switches into the local speech recognition mode if the network state is bad, to periodically monitor the network state in the local speech recognition mode.
|