[AG-TECH] Real Time Music Collaboration

Osland, CD (Chris) C.D.Osland at rl.ac.uk
Fri Mar 8 11:35:03 CST 2002

A few techie thoughts.

The BBC tried this - I think it was on the occasion of the last
eclipse to pass over UK.  They had about 50 sites lined up to
do segments of a song - it might have been "A Beautiful Day" or
that might have been a different project.  They used ISDN for
the links.  There's an article about it in Audio Media - I'll
try to dig out a URL.

As far as I remember 'n' sites required either 2n or 4n delay
units.  I think there was a central source (aka click track
with something recognizable on it beyond just clicks) which was
radiated out to everyone.  They did their bit in time with the
click track.  All the feeds were then relayed back to a central
site, put through a site-specific delay unit, and combined.
This was actually done on TV, not radio, which made things
even more difficult, in terms of producing a network feed.

What happened was pretty well chaos and confusion, which was not
too surprising considering that there was little time to test.

I think two main problems caused the chaos.  The first was that
although someone had tuned up the delays so that the round trip
to each site and back was the same (i.e. at least as long as the
longest) they didn't realise that these would vary over time.
By the time of the live event, various links had dropped and
had to be re-established and the round trip time was different.
I think this was the minor problem.

The more crucial thing was that people did not realise the
vital need to hear the click track when everything got
going - a bit like having on-stage monitors so that a band
stays together.

At a guess things were probably made worse by people trying to
listen to the *broadcast*, not the *click track*, and this
was probably a second or so later than them, having gone back
to the central studio, gone through a delay, been mixed and
munged and sent to the transmitter, etc..

I don't know what the delays are through an AG system, but I
bet they vary from site to site and are above 20 ms!  I suspect
that collaboration will therefore require some approach
similar to the above, where players hear a central source
but mute the other collaborators.  I can't remember whether
one can specify a delay in RAT - I think this will be
needed on a site-by-site basis.  The other possibility would
be for site 1 to play and be listened to and added to by site
2 which is listened to and added to by site 3 ...  This
would be a real-time equivalent of a multi-layer build-up
of a number, just with no retakes!

In both cases everyone involved would only be able to
hear the finished result from a recording, not while
they were playing, although in the latter scenario
people towards the end of the chain would hear an
almost complete version.

As mentioned in my e-mail about our band's membership, I
have facilities for multitrack recording that I can bring
in from my studio to work (DSL is not really a concept in
the UK so I won't even think about doing this from home)
but would be interested to know whether anyone knows of a
variant of RAT that would feed each sites audio to a
separate output.  I could probably configure a couple
of Korg 1212 I/O cards into our audio machine which would
give me 24 ins and outs.

As you'll gather, I'm excited about the possibilities of
this - I'm just off to find Linux drivers for the Korg 1212!



Chris Osland                         Office tel: +44 (0) 1235 446565
Digital Media and Access Grid      Medialab tel: +44 (0) 1235 446459
BIT Department             Access Grid room tel: +44 (0) 1235 445666
e-mail:   C.D.Osland at rl.ac.uk               Fax: +44 (0) 1235 445597

CLRC Rutherford Appleton Laboratory (Bldg. R18)
Chilton, DIDCOT, Oxon OX11 0QX, UK

[The contents of this email are confidential and
are for the use of the intended recipient only.
If you are not the intended recipient do not take
any action on it or show it to anyone else,
but return this email to the sender and delete your copy of it.]

-----Original Message-----
From: Jonathan C. Humfrey [mailto:jch at cs.ucsb.edu]
Sent: 07 March 2002 18:20
To: jch at cs.ucsb.edu
Cc: ag-tech at mcs.anl.gov
Subject: [AG-TECH] Real Time Music Collaboration

I am looking for someone to do a real time music collaboration experiment
over the AG.  Is anyone interested?

Jonathan Humfrey

More information about the ag-tech mailing list