Development of Multi-site Voice Control System for Space Collaboration System

本研究的目的是开发一种语音控制系统，该系统可以自动安排语音电平，以便在大学间卫星网络(SCS)的使用中进行顺利的讨论。虽然SCS中的大部分视听设备都是自动控制的，操作方便，但由于自动化的困难，用户不得不进行音量安排。为了安排每个站点的语音级别以保持整个系统的平衡，一种解决方案被认为是在站点之间共享关于语音级别的信息。然后，我们开发了一个在卫星网络中的数据共享系统，并对其在南海VSAT站点上的实现进行了性能研究。为了利用SCS的功能，我们将重点放在了会话期间的材料传输上，并设计了与链路控制系统顺畅连接的系统。编解码器的信号多路复用功能使选择链路速度带宽成为可能。为了避免动态路径设置的复杂性，我们还研究了非对称闭合卫星网络，识别语音和噪声之间的差异是自动语音电平控制的必要条件，它不仅可以抑制过多的输入，而且可以将小的语音提高到适当的水平。对SCS中使用的声学环境的测量表明，声级的阈值设置不足以进行声级安排。研究了具有频率依赖性输入输出特性的声级控制系统，重点研究了语音和噪声之间的频率特性差异。

The research purpose is to develop a voice control system which automatically arranges a voice level, so that one can make smooth discussion in the use of the interuniversity satellite network (SCS). Although most of the audio-visual equipments are automatically controlled with easy operation in the SCS, users had to make the voice level arrangement because of difficulty of the automation. To arrange the voice level of each station keeping balance of the whole system, one solution is considered sharing information on the voice level between stations. Then, we developed a data sharing system in the satellite network, quid studied on the performance implementing it into VSAT stations of the SCS.The developed data sharing system allows us to make a multicast file transfer. To exploit the SCS features, we focused on material transmission during a session, and designed the system with smooth connectivity to the link control system. Signal multiplex function of a CODEC makes it possible to select a link-speed bandwidth. We also investigated on the asymmetric-closed satellite network in order to avoid complexity of dynamic path setting.Recognition of difference between voice and noise is necessary for automatic voice level control that not only suppresses excessive inputs, but also gains small voice to proper level. The measurement of the acoustic environment used in the SCS indicated that threshold setting of the sound level is not enough for voice level arrangement. We studied on the voice level control system that has input-output property with frequency dependency, focusing on frequency property difference between voice and noise.