Distributed speech recognition system and method

A distributed speech recognition system and method according to the invention, to the input signal to be input by using the Detection of quiet periods speech interval enables the word recognition and natural language recognition, the speech recognition target required by the various terminals of various because by using the identifier of the terminal for different recognition vocabulary group (e.g., domestic voice recognition vocabulary group, a vehicle telematics vocabulary group call vocabulary group and so on for center) in the same speech recognition system to screen the recognition vocabulary, the group to which that terminal required It will have to be processed. Further, by minimizing adapted to a voice database model the effects of various channel distortion according to the terminal type and the recognition environment to the channel estimation method is one to improve the speech recognition performance.