Content-Based Audio Retrieval Using a Generalized Algorithm

Abstract

Content-based indexing of audio (and multimedia) data has become more important since conventional databases cannot provide the necessary efficiency and performance [1,2]. However, there are three main difficult problems. First, the content of audio data is subjective information; it is hard to give the descriptions in words. The recognition of data content requires prior knowledge and special techniques in Signal Processing and Pattern Recognition, which usually require long computing time. Second, since several audio features can be used as indices [3] (such as pitch, amplitude, and frequency), a method or processing technique designed and developed for one feature may not be appropriate for another. Third, the extremely large data size and the use of a similarity search require extensive computation. Similarity matching is based upon the computation of the distance between a query and each record in the database; the best match is in the data set with the smallest distances. To solve these three problems, we use a histogram-based feature model to represent subjective features[4], a unified model [5] to represent the data structures of the multimedia data, and a fast, generalized comparison algorithm to reduce the retrieval time.