Re: CU-SeeMe - Does Audio work on a Modem?

Jason Williams (streak@ccwf.cc.utexas.edu)
Tue, 6 Jan 1998 13:06:48 -0600 (CST)


On Tue, 6 Jan 1998, Gary Dietz wrote:
> First, let me say that with this environment: Pent 166, 32Mb RAM, 33.6
> modem, my local ISP (Empire.Net), and CU-SeeMe 3.1 (released version) I've
> achieved phone voice quality point-to-point calls on a regular basis using
> the new G.723 audio codec.

I always assumed the more bandwidth an audio codec consumed, the better
the quality (and the less compression and hence less CPU power required).
If that's true, then the Delta-mod codec would be better quality than
G.723..True? I can't confirm since I can't get any audio to work at all
thru CU 3.1. G.723 work very well for music broadcasts?

> So why the mixed results? Why the irritating choppiness?

I've had some interesting experiences with audio that I don't quite
understand. Using the 2.1.2 version (the only WP version I can get audio
to work with), the quality of a single audio codec depended on what
sending codec I had it set to. The sound coming from other participants
when I had it set to Digitalk was drastically different than when I had it
set to Delta-Mod..even when they were sending the same codec all the time.

> Well, last night, I was at work (on our Lan/T1) and called my Mom on CU 3.1
> through her AOL account. She has a Pent 100, 24Mb of RAM, and AOL and a
> 33.6 modem. The audio REALLY sucked. (But I was getting her MJPEG video
> okay--though slowly.) Why?

<gasp>..You mean your mom's CU world isn't perfect? </sarcasm> :)

> Well, at 8:00PM EST, she dialed into AOL and got a "24KBPS" connection.
> Also, from my POV, her statistics showed a 70% packet loss and an actual
> throughput of only *9*Kbps. It was a wonder we could even see each other!

Traceroute works wonders :) It's a wonder she hasn't experienced that
before though.

> a. Just because you have a 33.6 modem doesn't mean you have a 33.6 connection

Yep..until bandwidth reservation happens along...you live at the mercy of
the routers and machines that carry your traffic.

> b. Your send a receive settings must be set correctly. See the
> documentation or the White Pine Support pages.

I had a question about this...Not sure if you can answer it..But does the
sending rate limit the codec? ie: If you have your max-send set to 10kbps
and you try to send 16kbps delta-mod, does it get clipped? Or does it try
to send the full 16kbps if there's no packet loss? I can understand
trying to receive audio with bad rates, but does the max send also effect
audio? I've always set my max send up to 20 or so then paused for
Delta-mod.

> c. Don't expect to open 12 video windows, a whiteboard, and get audio on a
> 33.6 connection. It ain't gonna happen! Or, if it does, it will be very
> slow.

I wouldn't expect to open the whiteboard at all on a modem :)

> d. AOL and other ISPs are often oversubscribed for evening hours and it is
> very hard to get enough bandwidth for even minimum requirements for
> Internet Conferencing.

Now if only there were stats pages so you can tell WHICH ISPs are
oversubscribed.

> f. On a "public" server, you never know who is sending what or how they are
> sending it. If you run (or rent time) on a private server, you'll usually
> have better group conferencing results. And, with security and parental
> control needs, it is much better to purhcase your own server, or go to
> Powerscourt to rent time on a server.

Nice advertising :)
I don't quite see how conferencing on a private server is any different
than conferencing on a fairly unused public server with respect to the
quality of transmission and reception of audio. It's just as easy for
someone to have screwed up rates on a private server as it is for a public
server (After running both public and private reflectors..I can vouch for
that)
Of course...with programs like Refmarshal out there, more and more
reflectors (both White Pine and Cornell/Enhanced Reflectors) can be
controlled with a nicely designed interface. The issue of monitoring and
controlling will eventually be a thing of the past with more reflector
operators taking advantage of the tools out there.

> i. Finally, CU-SeeMe 3.1 has improved audio subsystems, and you'll get even
> better results even if you just use the Voxware codec.

improved? I'll let you know if and when I can get it to work :)
Also keep in mind, not EVERYONE uses CU 3.1...so sending G.723 audio to a
non-3.1 user has little effect. It's why a lot of people are still
sending in B&W from what I've seen...backwards compatibility.

--
streak@ccwf.cc.utexas.edu    * Jason Williams -- Austin, Tx.  |     |
streak@mail.utexas.edu       * University of Texas at Austin  | ___ |
streak@cs.utexas.edu         * BS Computer Science             \_|_/
*************** http://ccwf.cc.utexas.edu/~streak/ **************|