Design Of Audio Publications Automatic Segmentation System

With scientific and technological development, audio technology of publications has become a trend. Character segmentation and voice segmentation is the two main problems in the process of audio publications. In the current product design process, these two technologies are done manually. To solve these problems, the paper have studied the techniques of automatic segmentation system and realized it, which greatly improved the efficiency and accuracy.In this paper, compared with the classical method of character segmentation and speech segmentation, we has improved the traditional method and applied to automatic segmentation system. Finally, we realized the automatic segmentation of the character and the speech. The main research work of this paper is as follow: (1) In the document page segmentation part, based on the research and experimentation of the existing algorithms, we choose the algorithm based on pattern spread analysis to complete the pattern search and classification. Then the graphics area and the text area are separately exported. (2) In the character segmentation part, we proposed a width-based character segmentation algorithm to segment the Chinese characters and a interval-based segmentation algorithm to segment the English characters. (3) In the speech segmentation part, we proposed an adaptive threshold segmentation model to split the audio files effectively. (4) The automatic segmentation system is realized on Matlab platform. First of all the system is able to output the text area of the image, the character segmentation result is exported. Secondly, the system can complete the analysis of documents and automatic speech segmentation.In this paper, experiment of the character and speech segmentation algorithm proofed the feasibility of the system. At the same time, we analysed the causes of errors, and proposed suggestions to improve the performance of the system. Experimental data show that the character segmentation algorithm proposed in this paper and the principles of the speech segmentation algorithm is correct and the algorithms are technically feasible. The system has an important referential significance for the relevant product development.