Program climax search method and system

Abstract

The invention provides a climax search system and a climax search method for searching a program climax, and a corresponding computer product. The climax search system comprises a sliding window characteristic extractor, a model classifier, a frequency spectrum energy analyzer and a boundary finder, wherein the sliding window characteristic extractor is used for windowing an audio stream to be processed and extracting characteristic information of audio stream fragments in a window; the model classifier is used for scoring each audio stream fragment on a hybrid Gaussian model so as to judge a climax fragment; the frequency spectrum energy analyzer is used for analyzing frequency spectrum sub-band energy of the searched climax fragment and background environment noise and calculating an average energy value of the climax fragment; the average energy value is taken as a frequency spectrum sub-band energy threshold value; and the boundary finder is used for forwardly and backwardly calculating the sub-band average energy of the boundary position of the searched climax fragment respectively and finding a position on which the average energy is lower than the frequency spectrum sub-band energy threshold value and taking the position as the actual boundary of the climax fragment so as to search an entire climax fragment.

Description

technical field [0001] The invention relates to an audio retrieval method and system, in particular to a program climax retrieval method and system for retrieving program climaxes in sports game programs. Background technique [0002] Today's world is in the era of information explosion, and information is increasing at an exponential rate. The continuous development of multimedia technology and Internet technology has greatly increased the necessity of automatic analysis and processing of massive multimedia data. However, video analysis has a large amount of computation and consumes more resources, so audio analysis of multimedia data has greater advantages. [0003] Generally speaking, videos such as sports games are relatively long, and the content that sports fans really care about usually only accounts for a very small part of the entire content. If you need to find the content you care about, you often need to traverse from the beginning to the end to find the desire...

AI Technical Summary

Problems solved by technology

[0012] It can be seen from the above literature that the existing technologies only analyze and retrieve the specific content of a certain two types of sports competition content. These technologies are highly targeted and cannot be well extended to content retrieval for other categories of extracted competitions.

Images