A Design And Implementation Of Social Security Voice Service System Based On CTI

At present, although social security voice service system has been put into use inprefecture-level and provincial capital cities, it is not much in county-level cities. Inbig cities, plans of the system are always complicated and enjoy a large investment ofhardware because of large amount of call hits. Based on the analysis of call hits ofthese cities’ social security systems, and according to their characteristics of smallscale that numbers of manual seats and auto-answer seats are both less than10, thisthesis designs a voice service system with low cost of development that suitscounty-level cities, in combination with the plans of big cities.First, this thesis makes comparisons and analysis of the construction mode andscheme of existing voice service system. In the design of the system, according to thescale of the system in county-level cities, we combine Asterisk—the open source IPPrivate Branch Exchange software, with the voice card. The system contains manualseat and auto answer service and the article provides a comprehensive description ofthe design and realization of the interactive voice response module. The system usesan IP Private Branch Exchange server to achieve the centralized dispatcher of accessservices and the centralized management of resources. The system uses multi-threadtechnology and by VC++programming, it achieves communication between voicecard and ORACLE database and the relevance between voice service system andsocial security system, so users can get access to24-hour voice self service. The thesis also studies on the basic structure of the text-to-voice conversion system and build acorpus based on phrases combining the systems’ needs on digital voice synthesis. Thesystem achieves text standardization by adopting RMB conversion algorithm from thecapital form to the ordinary form and word segmentation by the forward maximummatching algorithm, and uses formant synthesis on synthetic voice. Finally,Experiences have proven that the intelligibility and articulation of synthetic voice aresatisfactory; characters are read naturally, so as to improve the voice interactiveresponse ability of this system.