APPENDIX B PARAMETRES OF the MODELS
- 格式:pdf
- 大小:334.91 KB
- 文档页数:20
Dalle Molle Institute for Perceptual Artificial Intelligence • P.O.Box 592 • Martigny • Valais • Switzerland
DECEMBER 2001
Speech Recognition Engine for Interactive Voice Response application on Windows
Haiyan Wang IDIAP-Com 01-10
phone +41 – 27 – 721 77 11 fax +41 – 27 – 721 77 12 e-mail secretariat@idiap.chinternet http://www.idiap.ch
1. Introduction 1 2. Speech Recognition Engine........................................................................................2 2.1 Development Environment.......................................................................................3 2.2 Technology Requirements........................................................................................3 2.2.1 Technology Software..........................................................................................................4 2.2.2 Technology Models............................................................................................................6
3. Development and implementation..............................................................................7 3.1 Porting from Unix to PC..........................................................................................7 3.1.1 Data formats.......................................................................................................................7 3.1.2 Static constants name..........................................................................................................8 3.1.3 Function name....................................................................................................................8 3.1.4 Calling method...................................................................................................................8 3.1.5 Missing functions...............................................................................................................9 3.1.6 Old bugs..............................................................................................................................9
3.2 Integrating and Optimising code on Windows NT................................................10 3.2.1 Exchange of data through memory buffer........................................................................10 3.2.2 Using pipes to integrate modules......................................................................................10 3.2.3 Endeavors with Multi-thread............................................................................................11
4. Socket Programming................................................................................................12 5. Tests and Evaluations...............................................................................................12 5.1 Software Testing.....................................................................................................12 5.2 SRE Evaluation......................................................................................................13 6. Conclusion................................................................................................................13 7. Acknowledgement.....................................................................................................14 8. References.................................................................................................................15
1. Introduction This report presents a part work of the InfoVOX project. The work was to implement a Speech Recognition Engine (SRE) on Windows with state-of-the-art Speech Recognition (SR) technologies, and integrate SRE within an Interactive Voice Response (IVR) system.
With the increased use of cellular telephones and more and more government regulation of how and when you can use your cell phone, Speech Recognition is a powerful tool for many of today’s automated telephone services. The recent emergence of accurate large vocabulary recognition at acceptable cost has increased the range of practical applications. Research in this field has given us a deeper insight into the nature of speech (Articulation, Acoustic, Linguistic and Perceptual Descriptions) and the way we use it to communicate. The related fields of Computational Linguistics and Natural Language Processing have also contributed to this endeavor.
The main scientific goal of the InfoVOX project is to perform further research and development in the field of interactive voice servers, with applications in the key area of call centers for computer telephony applications. Moreover, the generic goal of the InfoVOX project is to improve the state-of-the-art automatic speech recognition and natural language processing (dialog) technologies to access large and complex databases, and to integrate this technology in a well defined computer telephony application prototype.