摘要 |
The present invention relates to an apparatus and a method of processing voice for classifying a sasang constitution and identifying a user. The present invention includes a feature vector extraction part which extracts a feature vector which includes the time information of a frame from a collected speech signal, a likelihood probability calculation part which calculates the likelihood probability from a classic model according to a constitution and a speaker by using the extracted feature vector, and a recognition and determination part which recognizes and determines user information based on the calculated likelihood probability. The time information of the frame is information of identifying the position of the frame in the total interval of the speech signal. [Reference numerals] (200) Voice processing device; (210) Feature vector extracting part; (220) Likelihood probability calculation part; (230) Constitution and speaker classic model storage part; (240) Recognition and determination part; (AA) Voice signal; (BB) Speaker recognizing result of constitution determining result |