SPEAKER RECOGNITION AND SPEECH RECOGNITION APPARATUS AND METHOD THEREOF

摘要

PURPOSE: A speaker/voice recognition apparatus and a method thereof are provided to improve convenience in use compared with text-dependent speaker recognition by automatically producing a speaker model according to confirmation results of a voice signal. CONSTITUTION: A feature vector extractor(121) extracts a feature vector from a voice signal of an input part(110). A speaker recognition part(122) recognizes a speaker based on the feature vector. When the voice signal is not a pre-registered speaker, the speaker recognition part produces a speaker model based on the feature vector. A voice recognition server(150) recognizes the voice by the voice model according to the confirmation results of the input voice signal.