Design And Implementation Of Video Caption Algorithm Based On Proposed Voice Activity Detection

With the rapid development of the Internet,people's daily lives have been changed tremendously.From the aged newspaper to the recent digital media,there are various formats of media.However,there exists a large number of people,such as the hearing-impaired,who are unable to hear the voice of videos.There is a gap between them and normal people when they are accessing information.Therefore,an alternative form to achieve the audio information in a video,such as captions are needed by the hearing-impaired group.This paper aims to solve the ubiquitous problem of the online videos' shortage of captions.Firstly,the article suggests to crawl online video contents and then converts them to audio contents,and it puts forward a new voice endpoint detectionWeevaluat.the effectiveness and feasibility of the proposed voice activity detection based on the Gammatone filter through experiments.Secondly,this article generates correction captions based on correction strategies so that captions not only become more suitable for the hearing-impaired but also assist the group to learn lip language.Concerning the hearing-impaired people's needs,this paper launches the research to discuss the problem of captions.According to the demand,it is suggested to crawl the corresponding online videos to build a service platform for the hearing-impaired people which provides the services of videos and captions.In the aspect of system implementation,we adopt Tomcat as the server and utilize the front end and background framework.Besides,we carry out the speech recognition and the caption adjustment.As a result,the videos and captions are displayed on the front page.