METHOD FOR EXTRACTING FEATURE VECTOR OF AUDIO DATA AND METHOD FOR MATCHING THE AUDIO DATA USING THE METHOD

摘要

PURPOSE: A feature vector extracting method of audio data and a matching method of audio data using the same are provided to accomplish high speed extraction and exact matching by changing various forms of sound sources to the same form and synchronizing the start point of music. CONSTITUTION: Audio data of the starting part of sound data is inputted. The audio data is decoded(S12). A start point where an audio is played from the decoded data is searched(S13~S15). Previous data of the searched start point is removed(S16). Feature vectors are extracted by processing data after the start point(S17). The extracted feature vectors are compared with feature vectors of audio data which are already stored in a database.