some Mac CUSM audio considerations

Geoff Rehn (EST)
Sat, 3 Dec 1994 13:59:15 -0500

At 10:37 AM 12/3/94 -0500, Dan Updegrove wrote:

Re: CU-SeeMe featured at CAUSE94 Conference

>Most in the audience apparently had never seen CU-SeeMe, so you might be
>hearing soon from university colleagues about their interest. (And, if so,
>you might do well to advise them that they couldn't count on Internet
>audio to perform as flawlessly for ten minutes in mid morning! I'm sure
>Glenn would have explained this if he'd had more time.)

The above observation leads me to humbly offer the following:

What is a method for ensuring reliable, audible and recognisable speech in
Macintosh CUSM videoconferencing?

I have found the following can be useful when using either Maven or
CU-SeeMe audio, _especially_ where bandwidth usage is a major
consideration, which for me is all the time. Very rarely have I been able
to indulge in any overall bandwidth greater than a total of 40kbps (video
and audio).

0. Before hand, get agreement amongst all parties on the audio protocol and
settings to use; have agreement on video CAP also, after some consideration
of the various bandwidth connections.

1. At the least, pause your own video transmission when talking yourself.

At the extreme, try getting _all_ parties to PAUSE sending video when
_anyone_ wants to engage in worthwhile audio interaction.

Either you can stop receiving the other party (ies) or the other party
(ies) can pause their video transmission.

Freeze one's own video shot to an "attractive" frame before pausing video,
ready to speak.

(Question: is pausing one's video _identical_ to stopping send, in terms of
bandwidth usage?)

2.Recommended: Use the latest Cu-SeeMe 0.70b14 or Maven2.0a18 (or 2,0d21)
with the new delta-mod encoding, at 16kbps - especially useful in low
bandwidth situations.

(I have found that Maven 16kbps alone, with no Cu-SeeMe open, gives great
audio over 28.8kbps SLIP dialup)

3. Using "Push to Talk" at all sites seems to minimise "feedback" problems
from the other site, where one's audio received at the other end(s) feeding
back into the other party's microphone, and hence back to you, especially
if the microphones are close to the Mac's speakers, as often they are.

I have found that on some Macs, CU-SeeMe "Push to Talk" won't work but
Maven "Push to Talk" will. I then use Maven for audio output, and open
CU-SeeMe with its audio window closed. (In this case, I am not sure which
application - Cu-SeeMe or Maven - receives the incoming audio).

4. Turning off "self-reflect" on the reflector being used, cuts feedback of
oneself while oneself is talking (!). (I guess Mute Speaker while talking
would do the same , but this is tedious, and one can forget it is muted).

Using a reflector with Maven enabled also allows for the widest combination
of possible participants - CuSeeMe video lurkers (those without a video
card etc) can use Maven to transmit audio and CUSM to receive video.

5. When connected by SLIP, I usually run Peter Lewis's Talk program
"pulsing" in the background, to keep the SLIP connection open.

I _suspect_ that Talk may (sometimes) interfere with Maven and possibly
CU-SeeMe audio. If I remember to, I turn Talk off, when SLIP connected.

If all parties follow the above, audio will be good, or at least have a
better chance of being so; it will also utilise the least bandwidth - a
consideration which is especially vital to getting the greatest possible
interactivity, enjoyment and quality out of (low bandwidth) desktop
videoconferencing, where one or many (if not all) parties have low
bandwidth connections.

For many of us, it may be perhaps too much to expect both video and audio
to be simultaneously high quality.

PS. I will be trying to put the above principle in practice tomorrow, in a
link with Japan and Canberra. I believe the Japan link will be only 32kps.
There will be three sites.

It may be best to then have only one site _at a time_ transmit either only
video or only audio (not both together); if we all CAP video at say 30kbps
and use 16kbps audio, we should all be able to see and hear each other
reasonably well.

I'll let you know...

Geoff R.

Geoff Rehn
Academic Services Unit
Murdoch University    Western Australia