Design And Realization Of Online Audiobook System

Reading has been an innate behavior of human beings since ancient times.Through reading,we can obtain information and understand the world,and can also relax and cultivate our sentiments.With the rapid development of the Internet,many people’s reading objects have gradually changed from paper books to e-books.However,the accelerated pace of life has led to increasingly fragmented time,and it is difficult for some people to spare time for reading.On the other hand,most novel platforms currently use a paid model,and the cost for users to read a book completely is high,so some users turn to pirated resources of poor quality.Not only that,long-term use of electronic screen reading will also lead to the decline of vision.In terms of search algorithms,traditional book search is implemented based on some search engines,and it is difficult to dig out the potential semantic connections between books in this way.Based on the above problems,this paper designs and implements an online listening book system.The development and implementation of this system mainly have the following four meanings: First,the emergence of the online listening book system can attract users to support the genuine version and improve the enthusiasm of the creators.It also improves the quality of books.Secondly,the way of listening to books enables users to obtain book information even when they are away from the screen,making full use of fragmented time and protecting eyesight.Third,processing book content through text dubbing and music can improve people’s reading interest and stimulate people’s desire to read.Finally,listening to books allows low-education groups such as children to access information without knowing the text.In the system introduced in this article,the client part is developed based on i OS,one of the mainstream systems of mobile devices,and the server part uses the SSM framework,My SQL database and Redis cache.The system is mainly divided into seven functional modules including book city,bookshelf,books,audio playback,resource search,community square and personal center.Among them,the bookstore and bookshelf modules are used to display simple information about recommended and collected books to users.The specific content information is displayed in the book module.The audio playback module is the core module of the system and is used to provide users with the function of listening to books.It provides the ability to search in massive book data,the community square provides a platform for users to share and communicate,and the personal center allows users to manage some personal information.On the basis of traditional search engine-based,this paper adds a search method based on semantic similarity,which is generally divided into two stages: recall and sorting.In this paper,in order to quickly obtain the more relevant Top K data from a large number of book data in the recall stage,the Sentence-BERT model is used to complete the calculation and storage of the semantic vector related to each book offline.During the recall phase,retrieval will be done through Faiss for efficiency.For the K results obtained in the recall stage,the BERT model will be used in the sorting stage to perform a secondary calculation of semantic similarity and re-rank it to improve the search accuracy.After testing,after using the SBERT model to recall 200 books,the BERT secondary sorting takes about 485 milliseconds and the accuracy rate is 84.1%.This subject is based on my internship project,and the development and testing of each functional module has been completed.At present,the online listening system is in good condition and has achieved the expected goals and effects.