Vaudeville: A high performance, voice activated teleconferencing application

We present Vaudeville, a voice-activated, hands-free, ATM-based video confe rencing application. This system is scalable; although video bandwidth is n ormally a limiting factor in the number of conferences participants, the ba ndwidth attributed to the video is not a function of conference size. This is achieved through an automatic, distributed floor control mechanism that gives the appearance of an open floor. Audio and video are encoded in hardw are using a platform-independent, ATM hardware multimedia interface. Vaudev ille features digitally transmitted NTSC video, voice-activated audio trans mission, audio bridging of two audio streams, and voice-activated video swi tching. Multiple simultaneous multiparty conferences are supported. Users c an move freely among conferences without knowledge of the underlying networ k structure. We describe how Vaudeville was built using a component-based d istributed programming environment. We also describe the algorithms used to control the audio and video of the application.