Up: Specification of the MICE Conference Management and Multiplexing Centre
Next: Shared Workspace
Previous Page: Multiplexing in the compressed domain
Next Page: Shared Workspace
In principle, the distribution of audio to conference participants is somewhat simpler than the distribution of video, since audio compression and decompression is well within the capabilities of even low power workstations, and even (uncompressed) PCM audio at 64kb/s is within the capability of many networks. However, this situation is not quite as simple as may be assumed at first sight. In almost all conferences, audio is the single most important communication medium. It is more sensitive to delay than video (assuming that lip synchronisation is not an absolute priority) if two way communication is required, and is much more sensitive to jitter and packet loss than properly packetised video. This may be surprising at first, but in the face of high packet loss, video can compromise quality and resort to pure intra compression of pictures which (from a user's point of view) is relatively unaffected by packet loss. Audio has no such option. Recent tests of the Ebone UK-Paris and UK-Stockholm links show around 20-30% packet loss at 64kb/s, which results in largely incomprehensible audio.
Generally speaking, audio should be multicast directly to all receiving sites from the speaker's machine. Assuming the Mbone (Multicast Tunnel Backbone) has been configured to utilise the best quality links, regardless of their geographical routing, this will subject audio to the minimum delays and packet loss. There is less of a format conversion problem than with video, as most workstations can perform the decoding of whichever format the originator chooses in software.
However, the CMMC will not be completely unused for audio distribution for several reasons: