Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!ucbvax!agate!ziploc!eps
From: eps@toaster.SFSU.EDU (Eric P. Scott)
Newsgroups: comp.sys.next
Subject: Re: Talk though Ethernet
Message-ID: <1674@toaster.SFSU.EDU>
Date: 7 Jun 91 00:44:46 GMT
References: <1991Jun2.224148.20862@cec1.wustl.edu>
Reply-To: eps@cs.SFSU.EDU (Eric P. Scott)
Organization: San Francisco State University
Lines: 30

In article <1991Jun2.224148.20862@cec1.wustl.edu>
	wpcf@wucs1.wustl.edu (Pei Weng) writes:
>I am getting started doing some DSP programming on NeXT.What I am
>trying to do is:when I talked to my NeXT machine,another user on 
>the network can hear my talk and respond using his Next machine.
>Has anybody done similar things before? Any information will be
>greatly appreciated.Thanks in advance! 

The most promising work in this area (what will eventually become
the Internet standard for this) is a package called Voice
Terminal developed jointly by USC/ISI and BBN.  The current
version runs on SPARCstations (not NeXTs).  While limited
functionality is possible over UDP, a protocol specifically
designed for real-time applications (ST) is preferable.  This
means building a new SunOS kernel, or in the case of the NeXT,
implementing ST protocol as a loadable kernel server.  VT is
capable of four-way conferencing; it mixes mu-law audio from
multiple sources in real time.  It also places exceptional
demands on the machine; since ST frames are only 256 bytes
long, audio is nominally processed in 180-byte chunks (NeXT's
snd driver apparently only allows power-of-2-sized reads :-( ).
This means the CPU has to be able to handle microphone data
at least 45 times a second, as well as whatever comes in from
the network.  A SPARCstation 1 is said to be barely up to the
task (and it's useless for anything else while VT is running).

If you want a simple, not-quite-realtime talk, Adamation has a
program called Live Wire with voice capability.

					-=EPS=-