Font Size: a A A

Reasarch Of TTS/ASR In UMS

Posted on:2003-06-19Degree:MasterType:Thesis
Country:ChinaCandidate:C Q ChengFull Text:PDF
GTID:2168360065964216Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
This dissertation discusses important issues pertinent to the UMS (Unified Messaging System) based on the experience gained in the project of ZHX-UMS (ZhongHongXun Unified Messaging System). The work presented in this dissertation reflects the author's involvement in the development of the ZHX-UMS prototype system,and can be classified into two groups:system top-design and software development. An emphasis has been given to the software development,which directly related two techniques:TTS (Text-to-Speech) and ASR(Auto Speech Recognition).In chapter 2,a comprehensive discussion on the UMS is provided,which spans from the motivation of the UMS,its technical evolution,main features of existing system,through its architecture and market potentials. As a result of the discussion and analysis,a three-layer architecture is put forward as the basis of the ZHX-UMS development.In view that the UMS involves both text and voice oriented terminals,while all the information stored in a UMS is in a unified form,it has to deal with the conversion between the text formatted data (e.g.e-mail) and speech oriented data whenever input and output are needed. Consequently,a study on technique aspects of the TTS and the ASR is absolutely necessary,and fundamentals of TTS and ASR are discussed in chapter 3. Such discussion involves a wide spectrum of technique issues,such as functional structures and components,criteria for performance estimation,and the important issues relevant to the UMS.Chapters 4 and 5 deals with the development of communication software based on the TTS and ASR techniques with the help of the speech development tool kit,the Microsoft Speech SDK. In this work,object oriented approach is adopted,and basic modules are encapsulated as ActiveX control components to enable convenient communication among subsystems.The work presented in this dissertation is merely based on a ZHX-UMS prototype system,and the experiments so far carried out are limited,nevertheless,the preliminary results have shown that the basic design are workable and will surely provide a sound basis for future development.
Keywords/Search Tags:UMS(Unified Messaging System), Speech Technology, TTS(Text-To-Speech), ASR (Auto Speech Recognition)
PDF Full Text Request
Related items