SYSTEM FOR IDENTIFYING SPEAKER WITH UNFIXED CONTINUOUS NUMBERS SOUND AND METHOD THEREOF

摘要

PURPOSE: A system for identifying a speaker with unfixed continuous numbers sound and a method thereof are provided to improve a processing speed by using a simple number row, to prevent a surreptitious use of voice by increasing the voluntariness and to equip a mechanical voice reading mechanism. CONSTITUTION: A random number generating unit(240) generates unfixed continuous numbers. A user interface unit outputs the numbers to a user and receives a user's voice input about the numbers. A characteristic sampling unit(250) samples a voice characteristic of the received voice signal. A number sound identifying unit(220) samples a number corresponding to a matched item by comparing the characteristic with items of a speaker individual continuous number sound model. The number sound identifying unit(220) discriminates the sameness between the sampled/output numbers. A speaker identifying unit(230) measures the similarity by comparing the characteristic with a speaker model and decides a speaker identification based on the similarity.