Continuous Audio Stream Segmentation And Classification Systems Research

Audio signal processing and research on contemporary life is important. In this paper, a continuous audio stream for a number of structural features and related parameters were analyzed, detailing the application of clustering analysis and fuzzy cluster analysis to classify the design method to optimize the traditional fuzzy clustering analysis of the classical algorithm-FCM Algorithm parameters and determine the method, a modified fast clustering method and the experimental data for several methods in this experiment, it is proved that the algorithm is effective and reasonable. This research work mainly in the following areas:(1) analysis of information related to sample pretreatment methods and feature extraction methods, the typical method of determining the parameters described in detail, clearly the optimal feature and classifier design for the back to determine the relevant initialization information.(2) of the common audio sample instructions segmentation methods and elaboration of its structural parameters related to the definition and description, and the problems of traditional methods focus on analysis and classification Select a closely related fractal Method, and then use the data for comparative analysis of the experimental method for determining the accuracy.(3) a brief introduction of the common clustering and fuzzy clustering methods, and clustering analysis based on the objective function of a classical algorithm-FCM algorithm is described in detail, evaluate its effectiveness given the relevant mathematical definition of weighted index of the meaning of the algorithm are described, and its determination method are introduced. And on this basis proposes a spatial information to optimize the use of the direction of clustering methods, and audio samples are given for feature-based clustering algorithm improved space.(4) gives the division of the sample flow of information classification system design, mathematical models related to the above algorithm is applied to the sample information and the classification system design flow to address the common audio stream system can not extract information about a single classification and not Deterministic problem, the information flow for the extraction of the sample provides an effective method.