A Humming Retrieval System

How to analyze, store and retrieve the huge amount of data efficiently and effectively, especially for those multimedia data is an imperative problem. Because of the nonstructural character, the development of audio retrieval is restricted hugely. Comparing with image and video retrieval, audio retrieval study is behind hand. Content-based audio retrieval has been the studied hotspot of multimedia retrieval. This paper focuses on the key techniques of content-based audio retrieval, developed mainly in the following aspect.Through analyzing audio feature extraction and expression,This paper studies the audio physical feature, such as zero-crossing rate, short-time energy and power spectrum etc, and the audio perceptive features, such as loudness, brightness and pitch etc. Furthermore, this paper symbolizes loudness, brightness and pitch etc through notes, which can effectively express loudness, brightness and pitch.This paper studies content-based audio retrieval. We discuss the hidden Markov based sound retrieval, the fuzzy clustering based sound retrieval, and DTW algorithm based sound retrieval. Furthermore, this paper implements the fuzzy match which is based on the DTW algorithm.This paper implements a humming input and content-based sound retrieval prototype system which is fast, effect, no training needed, and can be well extended. In the end, this paper analyzes the result of the prototype system and compares them with those of already existed system.This paper prospects the developing trend and research hotspot. This paper proposes that on the one hand the trend of content-based music retrieval, and makes the extraction of perceptive features as the coming research orientation, and on the other hand, the analysis of sound field will become the main stream of research as well as the sound classifier which will also become the important research orientation.