【摘要】 <正> 1 语音混合简介视频会议系统中,多路语音混合分为5个阶段:1)多个终端通过IP网络向完成混合功能的服务器发送多媒体数据。由于是会议信息,其中包含语音、视频及控制、数据信息等。其中,音频和视频信息采用UDP作为传输协议,通过RTP将数据传送出去。而数据和控制信息则采用TCP作为传输协议。此时,多点信息是同时并行向混合端发送。2)服务器串行接收来自各用户的数据流,经解码后发往混合模块。3)混合模块串更多还原

【Abstract】 It is necessary for various users to intercommunicate with each other in multipoint communications. Each users’ image and voice can be reached at others’ terminals. As to the signs of image, many windows can be opened on a termined and the participarts’ images can be on the screen as well. However, as to voice, no matter how fast the speed of transmission is, it is common for listerners to find the great distinctions, between the voice of communications and that of face-to-face communications, which debases the quality of multipoint communications. Therefor, it is vital important to take measures to solve the mixed voices of multipoint communications during a high quality system of multipoint communications. In this article, we try to solve the trouble of voice in speech mixing. The theory of queue is applied in the analysis of voice data of multipoint communications, which forms the mathematic model. From the model, the equations of estate transfer will be obtained. And then through computer simulation, we get the results of equation and data we wanted are presented as well. With the model of queue, massive experiments and curvilinear regression, the relation among the scale of communication, the threshold of working, and the queue buffer is found. Therefore the expansibility problem of voice data of multipoint communications with real-time speech mixing is solved in the last part of this paper.更多还原