Audio is one of the most demanding media types and a stringent test of temporal precision on the communications medium. We present a novel architecture for delivering uncompressed multichannel audio over high-bandwidth ATM networks. Our architecture is robust, scalable, and can handle multiple audio channels. Furthermore, it does not impose a significantly overhead to the host CPU and allows multicasting to multiple participants, as well as quality of service support. We have implemented a high-speed local area ATM network on the USC campus that serves as the testbed for our experiments. A series of performance evaluations is presented to identify the effects of audio packet size, buffering, and network latency on the quality of multichannel program material.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.