Study On Speech Analysis System Based On Call Center

Call center is also be known as customer service center which has developed for more than ten years since it was introduced to China.As the medium between enterprise and customers communication,call center shoulders the common responsibility of monitoring service quality,improving the overall level degree of customer service.The development and popularization of call center in China improve rapidly,call center is the key to the communication between enterprise and customers,and it promotes the capacity of marketing information collection.The application of call center will help enterprise discover the marketing rule and improve competitiveness.In this research project,based on the technique of speech recognition and text analysis,with the help of enterprise frameworks of development,we independent research and develop the speech analysis system which consist of speech extraction program,single sign on module,speech recognition engine and text analysis engine.Through converting of speech into text,and the analysis of the text data,analysts can calculate the influence on the speed,sematic,motion of the speech content.On this basis,models can be established and the search can be implemented on the source data,also,the word frequency statistics,hot words statistics,call volume and models calculation can be achieved through the analysis system.The operator can evaluate the degree of the call center service and it’s more convenient to discover common mistakes during the calls,the results can be visualized by front-end interface.The research project adopts 100 hours corpus as the basic source data,the system gets a deep integration of Central Authentication Service framework which is used to achieve single sign on module.Through our labs preliminary work on speech recognition,we can obtain sufficient text data for the follow-up work.Dictionary segmentation algorithm is developed for text analysis,and Spring MVC,Angular JS Framework are also adapted for stable implementation,front-end methods are adapted for displaying the data result dynamic and vividly.