Інформаційна технологія пошуку аудіозаписів на основі kd-дерев

Abstract

The information technology for search of the audio recording by an audio fragment is developed. The model of audio corpus, containing an array of audiofiles, a database of parameters and metadata, dynamic kd-tree, is proposed, it has reduced the audio fragment duration, selected for the search There are suggested the new estimations of the measure proximity that take into account the specifics of the kd-tree use to find the nearest vector and the method of the combined search. The implementation of this method allows to achieve a significant reduction in the search time preserving its recall and precision.