Audio retrieval with fast relevance feedback based on constrained fuzzy clustering and stored index table

Abstract

Prior work in audio retrieval needs to generate audio templates by supervised learning and find similar audio clip based on pre-trained templates. This paper presents a new and efficient audio retrieval algorithm by unsupervised fuzzy clustering: first, audio features are extracted from compressed domain; second, these features are processed by temporal-spatial constrained fuzzy clustering, and the relevant audio clips can be represented by the clustering centroids; third, we use triangle tree to speedup the similarity measure. Relevance feedback is also implemented during retrieval. Therefore, the result can be adjusted according to users’ taste and is consistent with human perception.