Distributed Speech Recognition And Voice XML Standardlanguage In Vivid-Ring Application

Distributed Speech Recognition(DSR)is a newly emerging technique, which will be widely used in various areas. This paper makes a study of the architectures of Distributed Speech Recognitionsystems, realizes and evaluates a DSR system with Client / Server architectures. This DSR system USeS the speech recognition algo—rithm based on DDBHMM rDuration Distribution Based Hidden Markov Model). this system USeS multi—server and multi—client architecture and a load balance method to assign recogn ition tasks to diferent servers, and this makes the system have an excellent performance. In this paper the system structure of DSR is first introduced, then the major techniques used such as the front—end feature extraction algorithm, feature compression, error corection and detection, as well as the server side speech reconstruction algorithm are analyzed, and finally the potential application of DSR is briefly discussed.The vivid-ring development, according to the prototype of thETSIVR system architecture design and application of speech synthesis, speech recognition, and other key voice technology, including the use of voice xml standard language as a tool to develop a practical system : Users can call at any time through a Tebihao access to the system. button or through voice or songs tell the names of singers and interactive systems. you can convert their customized search CRBT, simple operation; For telephone process, mainly use the VXML to complete the development process.