| With the popularization and support of modern information technology,the development of distance education has been consolidated and strengthened,and the quality,sharing and fairness of educational resources have been realized.It is the pursuit of education at all levels of schools and ethnic areas in China.Due to the influence of transportation,geographical location and other factors,the educational resources of primary and secondary schools in China are relatively scarce compared with other regions.The characteristics of distance education and resource utilization in distance education make up for the geographical restrictions of Tibetan areas and expand the audience.Groups enable students in Tibetan areas to learn at any time and anywhere to realize national education informationization.The speech corpus in the distance education platform is an indispensable module for teaching resources,especially for minority language teaching resources.Among them,there are fewer teaching voice information resources in Tibetan middle schools,and the speech corpus is not enough to be targeted,and the phonetic labeling experts are also compared.Lack of a large number of Tibetan language teaching voice resources in primary and secondary schools is a very time-consuming and laborious task.In order to establish a high-quality speech corpus,in this paper,the TAN network-based annotation model is used to predict the accuracy of the labeler’s label data,so as to make the data with higher label trust degree into the corpus,and suppress the labeling of the difference labeler.The noise data enters the data set;it also enables the labeler to learn other people’s labeling experience by sharing data,improve the labeling level,and finally the labeling quality is controlled.This paper expounds the relevant principles of Bayesian network and Tibetan speech annotation system,and the necessity of application in distance education in middle schools in Tibetan areas.Manually labeling according to the recorded content,using MATLAB tool to construct the TAN network-based labeler evaluation model for the credential data of the labeler,controlling the voice corpus to accept only the voice data marked by the highly credential labeler,and realize the marker-assisted learning Model,build a high-quality Tibetan language middle school distance education speech corpus,,and then use Dreamweaver to realize Tibetan junior high school bilingual distance education website,strengthen Tibetan language learning in the distance learning process in Tibetan areas,improve learners’learning accuracy rate,and accelerate excellent educational resources.The spread of Tibetan areas promotes the process of modernization of education in Tibetan areas. |
