Based On Natural Language-specific Business Phone Voice Automated Advisory System

The touch-tone answering mode of human-computer interaction is widely used in domestic call center, which collects customer input by identifying telephone keypad. One hand, people feel more complicated, which will lead users to drop out or look for artificial service at the beginning. The other hand, some things are relatively complicated, so that customer input can not be collected by the touch-tone way. This paper studies natural language advisory system of the specific business based on telephone voice to get rid of the shackles of the telephone keypad. Using the natural language for human-computer interaction, as long as the user is on the phone to say the service, you can get the desired information. The method can improve the convenience of the query and greatly reduce costs. The system firstly accesses telephone voice signals with call center hardware platform, and converts the user’s voice input to text through voice recognition technology. Then understand the user’s query intent by the use of natural language processing techniques, and extract and query corresponding answer text from a business or FAQ database. Finally, use speech synthesis technology to get the answers voice and automatically answer by the call center platform. The specific work of this paper is in the following aspects.First of all, analyze the structure of typical call center, and discuss the basic principles of natural language advisory system of the specific business based on telephone voice, then design the overall structure of advisory system. The system includes call center answering platform, speech recognition, speech synthesis, answer extraction based on business and FAQ database. And research on how to build an efficient, stable and easily extended call center platform.Secondly with the secondary development interface of speech recognition, speech synthesis, which developed by the Institute of Acoustics, Chinese Academy of Sciences, to achieve speech recognition, speech synthesis of specific areas. Design a language model of easily extended rules to synchronously update the range of language identified by voice system with the contents of the database.Then research on answer acquisition method based on a business and FAQ database. For database query, analyzing the mapping relationship of defined query semantics block and database query language, achieve the conversion of natural language query to database query. For FAQ, extract answers from FAQ database by extracting elements of syntactic and calculating similarity of query statements.Finally, achieve the answering platform of call center with IPC plus voice board program, and then develop the smart dialing system based on telephone voice, self-check system of natural language flight information, FAQ search system of tourism domain.