SPEAKER IDENTIFICATION METHOD IN ARTIFICIAL INTELLIGENCE SECRETARIAL SERVICE IN WHICH CONTEXT-DEPENDENT SPEAKER IDENTIFICATION AND CONTEXT-INDEPENDENT SPEAKER IDENTIFICATION ARE CONVERGED, AND VOICE RECOGNITION DEVICE USED THEREFOR

摘要

Disclosed are a speaker identification method in an artificial intelligence secretarial service in which context-dependent speaker identification and context-independent speaker identification are converged, and a voice recognition device used therefor. The present invention is implemented through a process performed by a voice recognition device, the process comprising: storing, in a user's directory, user's voice utterance information relating to a predetermined call word; accumulatively storing, in the user's directory, voice utterance information of an unstructured natural language command following the user's call word utterance, by identifying the user on the basis of the user's call word voice utterance information; and generating a user voice parameter for context-independent speaker identification, on the basis of the voice utterance information of the unstructured natural language command accumulated and stored in the user's directory. According to the present invention, it is possible to prevent an authorized person from receiving an artificial intelligent secretarial service by using an artificial intelligence speaker.