|
||
Attempting to more accurately represent the process of audio loop collaboration on the network.
The transient series of events arose from the energies of Jesse Gilbert at CalArts, and has been taken along by the collection of people that have become involved starting from the CTL2000 laboratory in Adelaide. The format / structure is explained below. This information could be of use for people who want to partake, or people who want to set up their own transient loops. More information at some other Transient sites, one maintained by Jeremy Hicks, the other by Jesse Gilbert.
A collection of nodes in a loop. Each node is a small station with sound (so far) capabilities and streaming capabilities. Each node has a predecessor and a follower in the loop. The node takes in a stream from their predecessor, manipulates it in some way, then encodes again, passing it on. The expression "suck-spit" has been coined. The following node takes this stream and does the same thing. The sound stream thus travels from node to node until it arrives back at the "first" node, where it is again manipulated.
A central presentation node (or many such) can then take the collection of streams and present them to an audience. The technique used since the Transient II event at net.congestion is to use a loop of speakers and to pan the audio streams from one speaker to the next, keeping them separated. It is attempted to move the sounds such that audible events, that remain recognisable after manipulation at a node, should reappear at the same physical spot in the speaker array. Hardware and software to do this exists, but is not cheap. Jesse Gilbert at CalArts has a java setup and Times Up has some Max patches to control the Switchblade matrix mixer. Other hardware, such as digital, midi controllable mixers, could also be used.
At each node a stream is decoded, manipulated and encoded. The "classic" way to do this has been to use a realAudio stream, to have a machine decoding the stream from the predecessor, sound is then passed over to a mixer, effects units, tape players, whatever other manipulations seem appropriate. The resulting sound is then passed onto an encoder (often the same machine as the decoder, to be economical with hardware) and sent off to a realAudio server. It is better to use a remote server with good bandwidth connections. The producer is hidden at the bottom of this page.
A minimal system would be a machine capable of encoding and decoding realAudio, plus a mixer or effects device to play with the sound. A more complex system might use some effects, might add some layers of sound. It is desirable to leave a lot of the original stream in the produced stream, so that the manipulation process is gradual and subtle. Replacing the stream totally is not very interesting, as the central presentation place gets a collection of unrelated streams - a (boring?) cacophany - in any case, there is no loop or other interesting interaction going on, we may as well send cheap audio cassettes off to the central station ahead of time and save the bandwidth.
A recent development has been the practice of using a purely digital system; decoding, manipulating and encoding with no analogue pathways. An example of this is done using mp3 decoding, PD (Pure Data) manipulation and mp3 encoding. Information on the process, presently working only on Linux machines, can be found here.
It is advisable to try out hardware and software beforehand. Assemble a local loop - send an encoded stream to the server, decode it back, manipulate it and pass it back on. Such a local loop becomes a low-fidelity delay, but at least it tests the setup that you will be using. The transient list is a discussion forum for such events, announcements, discussions of times, people who want to be involved, hardware arrangements and other fun stuff take place.
Closer to the event time, a chat space is used to coordinate people. Previously we have used FMs telnet://chat.test.at as well as FogChat to do chatting in a two-dimensional space. As the participants come online, doing last minute tests and finding that certain nodes will not be able to take part, the order of the loop can be decided. Issues include the formats of encoding (RA players can play mp3, but someone needs separate machines to receive realAudio and encode mp3) as well as bandwidth and delays. If performances are fitting into a series of performances in an evening, the coordination of people, nodes and online-ness gets a bit hairy.
A suggested structure uses a sonic seed from one node, a small blurt of noise to get the loop jumpstarted. Keeping each node silent until the blurt arrives at the node, then manipulating it subtly, leads to a presentation in the central space where the sounds emanate from some place, then slowly scan around the loop of speakers. As the various nodes join in, the sounds remain separated but related by their common source - quite cool.
These events have been taking place for over a year now, sporadically but with non-waning interest.
One of the issues that we wanted to begin to look into but has not yet been touched upon more than briefly has been the addition of visual loop streams / manipulations. Methods for the presentation of such image streams have yet to be investigated, but this is slowly entering the realm of interest.
On this level, the presentation of the performance space, the net, the loop itself, has been started using the looped speakers. The presentation of the other main space, the chat rooms, has not really been looked into. FogChat is an interesting way to have chats that are spread in space, with colours as personal markers, so it becomes possible to follow who says what, when, in relation to what. The development of tools, whether java apps or more complex system dependent environments that _make_sense_ in some way, is a project that is harder and will remain so. For a while at least.
Research, as they say.
|
--->ctltheory Mailing List
|
--> CTL98 | --> CTL2000 | |||
 |
 |
||||