Feature Based Same Audio Perception method for Filtering of Illegal Music Contents

Recently, sharing and distribution of illegal music contents is increasing. For protecting that, Filtering have been using. Retrieval word interception of text searching and DRM are main technology in these days. Retrieval word interception of text searching method can not be used, if text information as a filename or tag information is changed. DRM is sublated because of usability limitation. We need a filtering that use audio data itself for these problems overcoming. In this paper, we proposed feature based same audio perception method for filtering of illegal music contents. We extracted feature data that have hearing characteristic from wave form data of music contents. Wave form of music file is always changed whenever it is resampled. We could get consistently feature data, although wave form wasn't perfect same. We compared each feature data and recognized identity by using Dynamic Time Warping algorithm. Recognizing results show that DTW is robust at time axis changes. We success all of 500 times experiment in randomly collected 1000 songs from same genre for proving of proposed feature based same audio perception method. 500 digital music files were made by mixing different compressing CODECS and sound qualities from 60 digital music. Feature based same audio perception method could find same music regardless of wave form changing. We proved that feature based same audio perception method is possible to filter illegal shared and distributed music contents.