Vat allows users to conduct host-to-host or multihost audio teleconferences
over an internet (multihost conferences require that the kernel support
IP multicast). On most systems, no hardware other than a microphone
is required -- sound I/O is via the built-in audio
hardware. On DEC systems, an AudioFile server must be running.
The conference is carried out on the address specified by
dest. Point-to-point conferences are initiated by supplying a
standard IP address, while multiparty conferences
use a Class D group address.
Port specifies the UDP ports to use,
and
ttl specifies the IP time-to-live (see below).
Port specifies two consecutive ports, one for data and
one for control. The data port is set to the greatest even
integer less than or equal to
port. The control port is one greater than the data port.
OPTIONS
(Note that all the parameters set by
the following flags can also be controlled by X resources
(which all have `reasonable' defaults)
so one should not need to give
vat any flags in the usual case. The flags only exist to temporarily
override some resource.)
-a
Enable automatic gain control on the output (speaker).
-A
Enable automatic gain control on the input (mike).
-c
Start up in `conference mode' (see description below). (This flag
is the opposite of
-l .)
-C
Use
conference as the title of this vat window. If no -C flag is given, the
destination address and port are used as the window title.
-d
Start up with the VU meters disabled (this can be changed using the
`Disable Meters' checkbox on the auxiliary controls panel).
-E
Include a checkbox in the auxiliary controls panel for specifying that
echo cancellation is being performed externally (i.e., in hardware).
This option can also be effected by setting the X resource
Vat.externalEchoCancel to ``true''.
-f
Sets the output audio packet format to
fmt. (Note that it not necessary to set an input format since vat will accept
any packet format it knows about.) Currently available audio formats are:
(The
dvi encoding was contributed by Jack Jansen of CWI.
The
gsm coder was contributed by Carsten Bormann
of the Technische Universitaet Berlin.
The
lpc coder was contributed by Ron Frederick of Xerox PARC.)
The default
audio format can be set with the
audioFormat X resource. It defaults to
pcm.
-g
Override the default window geometry with
geometry, which should be a standard X geometry string.
-I
Use the ``LBL Conference Bus'' to interact with other multimedia
conferencing tools.
The small integer
channel, which must be non-zero,
is used as the channel identifier for group interprocess communication on
the local host. This value should be consistent across each group
of applications that belong to a single conference, and should
be unique across conferences. The session directory tool (sd)
will allocate appropriate values.
(Vic and
vat currently use this mechanism to coordinate
voice-activated video switching as well as other
conference control abstractions.)
-j
Start up with audio output to the external audio jack.
This flag can be defaulted by setting the X resource
Vat.outputPort to
Jack.
-J
Start up with audio output muted.
-k
Keep all sites in the `Conference Hosts' panel. Normally sites that
exit are deleted from the panel. With
-k, sites that exit are grayed-out but not deleted.
-K
Use
key as the encryption key for this vat session.
-l
Start up in `lecture mode' (see description below).
This flag can be defaulted by setting the X resource
Vat.lectureMode to true.
-M
Start up with audio input unmuted.
-N
Use
session, in lieu of your user name and local host,
to identify you to other sites.
If no -N flag is given, the X resource Vat.sessionName is used.
-n
Use a packet format that is compatible with the obsolete
vat protocol, used by version 3 and earlier of vat.
-P
Use
priority as this vat window's priority for obtaining the audio device.
All vat windows have a priority which is typically set by the
X resource Vat.defaultPriority (defaults to 100) but this can be
overridden by a
-P flag. If a window requests the audio
(because new network data arrived or the mike has been unmuted)
and the window currently holding the audio is either lower priority,
the audio holder immediately gives it up. Otherwise the new window's
request is ignored.
-R
Disable the mike controls in the user-interface.
This prevents the user from enabling
the mike and transmitting audio to the session.
-s
Start up with audio output to the internal speaker.
(This flag is the opposite of
-j .)
-S
Make new sites come up `suppressed' (the check box next to the
sitename will be checked and you will have to click on it to hear
the site speak). This flag is intended for something like meeting
audiocasts where a moderator wants to have control over who is able to
speak. This flag can also be set by the Vat.muteNewSites X resource.
-t
Set the multicast ttl (time-to-live) to
ttl. (The ttl is ignored if the destination address is not an IP multicast
address.) If no -t flag is given, the value of the X resource
Vat.defaultTTL is used.
-U
Use the unix-domain stream socket specified by
socket for audio I/O. Some process should bind to and listen on
this socket before vat is run. The data is raw 8khz mulaw samples.
If
socket is a number, then AudioFile is used. The number indicates
the corresponding AudioFile device.
-u
Source
script, in addition to the compiled-in script, to build the user interface.
You can provide addtional tcl code that is loaded at
startup in $HOME/.vat.tcl.
-X
Override the X resource
Vat.resource
with
value.
VAT OPERATION
Note: In addition to invoking the ``quit'' button, typing `q', `Q', ctrl-C or
ctrl-D anywhere in the window will terminate
vat. The
vat window is divided into two parts: the right has controls for the local
audio and the left has a status display of the remote users participating
in the current conference. The audio controls consist of two sliders
that control the mike and playback gain, a button to toggle output
between the built in speaker and the headphone jack, buttons to
enable/disable either the mike or speaker, and buttons to control
acquistion of the audio hardware.
Just to the left of each
slider is a VU meter. A rule of thumb is to adjust the mike and
speaker gain sliders so the peak readings on the meter are about 80%
of full scale.
To change the audio output line (i.e., speaker, headphone, lineout, etc.)
click on the speaker icon (it should change to a headphone icon).
Additional clicks will round-robin among the available lines.
If there is only one option, the button will be disabled.
Similarly, click on the mike icon to select among the input lines.
By default, vat starts with the mike muted and the speaker enabled.
The mike is active when the ``talk'' button is selected, while
output can be muted by de-selecting the ``listen'' button.
The
Conference Hosts window lists all the remote users
participating in the conference.
Each user's name
is displayed in a box that is highlighted whenever that user
speaks and grayed-out if a `session' message from that
user hasn't been received for at least 30 seconds (vat sends
`session' message every so often) -- this usually indicates that
the site has lost connectivity or that vat has been aborted or
stopped. There is a checkbox to the left of each participant name.
Clicking on the box will cause audio from that participant to
be discarded instead of played (for example, this might be used to
suppress a site that is generating echoes).
Multiple VAT Windows
One host can be running an arbitrary number of vat sessions (presumably
with different destination addresses). However, since most workstations
have only one set of audio hardware, only one of those sessions
will be able to access the mike and speaker.
For the most part, the vat sessions will automatically follow the action.
If you select the ``talk'' button or press ``Keep Audio'' button, the audio
device will be acquired by that session and the session that
previously held the audio will relinquish it. Vat displays
it's title bar in an oblique font when the audio is not being held.
A vat session will also acquire the audio if there is input
from the network. But to prevent a background vat session
from stealing the audio from the foreground session, you
can toggle the ``Keep Audio'' button. When the ``Keep Audio''
button is selected, vat will reliquish the audio only if
there is a user demand in another window (i.e., unmuting the mike
or selecting the ``Keep Audio'' button).
Participants in a multi-site conference often want to have
`side conversations' that don't bother the rest of the conference
participants.
Vat has some support for establishing side conversations:
If you middle-click on the name of some
site in the conference hosts window, a new vat window will be created
that talks only to that participant (it sends unicast datagrams rather
than multicast). If that other participant also middle-clicks on your
site, you can have a private conversation between just your two sites
using the newly created vat windows. Note: due to a `bug' in
the way most systems implement multicast, if you create a new window
aimed at a particular participant but they haven't created a window
aimed at you, they will hear you speaking in the main conference window
and may not realize that your audio is being sent only to them and
not multicast. One can view this either as a feature (it provides
a semi-private channel you can use to ask someone to set up a side
conversation) or a bug (it often leads to strange, one-sided
conversations where one side multicasts and the other doesn't).
Auxiliary Controls
Clicking on the ``Menu'' label at the bottom of the
vat window will cause a panel of
auxiliary controls to open.
The
Audio Tests buttons will enable some audio test modes. These should
not be selected during a conference.
The
loopback mike button will cause input from the mike to be sent to the local
speaker/jack. This might be useful for checking levels and
debugging cable problems but the 20ms delay from input to output
makes talking in this mode almost impossible. The three tone
buttons will generate one of three reference tones through the local
speaker. Level setting should generally be done with the -6dBm
tone.
The
Output Mode buttons control what vat will do to avoid feedback/echo from the
mike to the speaker. In
mike mutes net mode, vat will mute the speaker whenever it thinks that you are talking,
while in
net mutes mike mode, vat will mute the mike whenever input from the network arrives.
In
full duplex mode, vat will assume that feedback can't happen and do nothing to avoid it.
In
echo cancel mode, vat will attempt to eliminate echoes by doing some fancy signal
processing. (EchoCancel requires the BSD sound driver -- it is
disabled when running vat under Sun OS because the Sun driver does not
provide any
mechanism to time correlate audio output and input.) The internal
speaker should only be used in `speakerphone' or `echo cancel' mode --
selecting `headphone' mode for it will result in your site injecting
a lot of unpleasant echoes into the conference. The headphone jack
should be set to `FullDuplex' mode if you have headphones plugged into
it and `MikeMutesNet' or `EchoCancel' mode if you have an external
amp and speaker plugged into it.
There are two
"type-in boxes" (see below) at the bottom of the Auxiliary Controls panel. The one labeled
`Name' can be used to change the session name announced to other sites.
The one labelled
`Key' can be used to specify an encryption key (see next section).
Encryption
(N.B.: Because of U.S. export controls,
the standard distribution of vat from LBL does not support encryption.
In this case, the ``Key'' type-in box will be disabled.)
Since vat conversations are typically conducted over open IP networks
there is no way to prevent eavesdropping, particularly for multicast
conferences. To add some measure of privacy, vat allows the audio
packet streams to be DES encrypted. Presumably only sites sharing
the same key will be able to decrypt and
listen to the encrypted audio.
Encryption is enabled by entering an arbitrary string in the
key box (this string is the previously agreed upon encryption key
for the conference - note that key distribution should be
done by mechanisms totally separate from vat). Encryption
can be turned off by entering a null string (just a carriage
return or any string starting with a blank) in the
key box.
X Resources
The following are the names and default values of X resources used by
vat. This list is incomplete. Consult the tcl code in ui-resource.tcl from
the vat source distribution for the complete set.
Vat.maxPlayout is the maximum `play out' delay, in seconds, that can be tolerated. I.e., vat dynamically adapts to delays introduced in the network by delaying
the play out of a remote site's audio packets. The range of adaptation
is limited by the size of a buffer in vat and this parameter essentially
sets the size of that buffer. Setting maxPlayout larger than 10 seconds will probably result poor vat and system behavior.
Vat has two different modes of adapting the playout delay, one more
suitable for an interactive, multi-party discussion or conference and
the other more suitable for listening to a speech or lecture.
The two modes differ in how quickly they `forget' the delay vat
introduces to adapt to transient network congestion:
In "Conference mode" vat attempts to minimize the delay (since large
delays make interactive conversations difficult) but this usually
results in more lost packets when the delay becomes too short handle
the next congestion event. In "Lecture mode" vat attempts to minimize lost packets by reducing delays very slowly.
This results in the clearest playback but interactivity may suffer.
"Conference mode" is the default when vat starts up unless the -l flag is given or the X resource lectureMode is set to true. There are radio buttons in the network section of
the Auxiliary Controls panel to switch between Conference and Lecture
Statistics
Clicking on a name with the left mouse button will bring up a
small window of identification information for that user.
The window includes several of the RTP identification descriptors,
the audio format in use, and the times of reception of the
last data and control packets.
A packet statistics window can be brought up from the ``Stats...''
pulldown menu by selecting ``RTP''.
There are three columns of numbers. The last column is
the aggregate statistics
since vat started, while the middle column is the difference between
the last update time and the current time.
The first column is an smoother version of the middle column, using
an exponentially weighted average with gain Vat.statTimeConst.
The statistics are
updated every second or so while the window is mapped.
Any of values can be plotted with a stripchart by clicking
on the name of the desired statistic.
Van Jacobson (van@ee.lbl.gov) and
Steven McCanne (mccanne@ee.lbl.gov), both of
Lawrence Berkeley Laboratory, University of California, Berkeley, CA.
Jack Jansen (Jack.Jansen@cwi.nl) of
Stichting Mathematisch Centrum, Amsterdam, the Netherlands,
contributed the Intel DVI ADPCM codec.
Ron Frederick (frederic@parc.xerox.com) of Xerox PARC, Palo Alto, CA,
contributed the LPC codec
which is based on an implementation done by Ron Zuckerman
(ronzu@isu.comm.mot.com) of Motorola which was posted to the Usenet
group comp.dsp on 26 June 1992.
Carsten Bormann (cabo@cs.tu-berlin.de) and
Jutta Degener (jutta@cs.tu-berlin.de)
of the Communications and Operating Systems Research Group (KBS) at the
Technische Universitaet Berlin
contributed the GSM codec.
Steve Casner (casner@isi.edu) of ISI,
Los Angeles, CA, and Steve Deering (deering@parc.xerox.com) of
Xerox PARC have invested tremendous effort in
making vat work on a scale far beyond the authors' wildest
expectations and have contributed greatly to vat's development,
both directly (via careful analysis of bugs and useful suggestions)
and indirectly (via setting up several global conferences that
severely pushed the envelope of vat's capabilities).
BUGS
Speakerphone mode is difficult to get right -- use a headset if you
can (or run BSD instead of Sun-OS to get a kernel audio driver that
can support echo cancellation). If you have to use speakerphone mode,
try to position the mike as far as possible from the speaker (the
speaker in a sparcstation is on bottom of the machine in the front
right corner near the LED). If there's a problem with echo (i.e.,
you transmit whenever other people start speaking), try reducing the mike
gain or mute the mike when you're not speaking.
In speakerphone mode vat assumes that if there is audio data from the net being sent to the
speaker at least part of the signal from the mike is pickup from
the speaker. So, unless the mike signal is `large' compared to the
signal from the net, vat assumes it is echo and suppresses it. This means that if you want
to interrupt someone who is talking, you may have to talk a bit
louder than usual at the start (you can tell if you succeed because
your site's name box will light and the speaker will mute).
