北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2007, Vol. 30 ›› Issue (1): 90-95.doi: 10.13190/jbupt.200701.90.wangwl

• 研究报告 • 上一篇    下一篇

VoiceXML语音平台中的语音资源预取算法

王文林, 廖建新, 朱晓民, 王纯   

  1. 北京邮电大学 网络与交换技术国家重点实验室, 北京 100876
  • 收稿日期:2006-01-01 修回日期:1900-01-01 出版日期:2007-03-30 发布日期:2007-03-30
  • 通讯作者: 王文林

A Voice Resource Prefetch Algorithm for the Voice Platform Based on VoiceXML

WANG Wen-lin, LIAO Jian-xin, ZHU Xiao-min, WANG Chun   

  1. State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China
  • Received:2006-01-01 Revised:1900-01-01 Online:2007-03-30 Published:2007-03-30
  • Contact: WANG Wen-lin

摘要:

在分析目前主要预取算法优劣的基础上,根据VoiceXML语音平台与基于HTML的WWW之间的区别,提出在VoiceXML语音平台中应该预取其引用的语音资源,在采用基于热点预取技术的同时提出一种自适应的多用户共享的Markov模型,可以统一预测所有在线用户下一步所需的资源及其访问概率,有助于提高预测的准确率.仿真研究表明,与单用户Markov预测模型相比较,这种多用户共享的Markov预测模型能在相同带宽消耗下得到更好的命中率,减少用户请求的访问延迟,提高响应速度.

关键词: VoiceXML, 预取, 多用户Markov预测模型

Abstract:

By analyzing the present dominating prefetch algorithms’ advantages and disadvantages, according to the difference between VoiceXML-based voice platform and HTML-based World Wide Web, it is proposed that the cited voice resource should be prefetched in VoiceXML-based voice platform. At the same time of adopting the hot-spot-based prefetch technology, an adaptive multi-user shared Markov prediction model is presented, which can predict the forthcoming required resource of all the online users and its probability to improve the accuracy of the prediction. The simulation research showed that this multi-user shared Markov prediction model could get a higher hit ratio, reduce delay of a user’s request and improve response speed comparing with the single-user Markov prediction model with the same consumed bandwidth.

Key words: VoiceXML, Prefetch, Multi-user shared Markov predict model

中图分类号: