Sounding the Net: Interview with Chris Chafe

Peter Traub: The SoundWire project seems directed at two things: creating a means by which QoS can be intuitively determined through listening to the network, and testing and creating an environment that allows for multichannel high quality real-time collaboration over long distances. From what I read about the latter direction, it seems much of the focus has been on making the network collaborations as transparent as possible, so that works and collaborations done over it aren't really about the network or state of the network, but just use it as a tool, like radio or TV.

Chris Chafe: ...good description of our intention to create distance musical environments keeping performance as "natural" as possible

PT: How do you think about the network in relation to composition and performance? i.e., a tool, a medium, a subject, etc...

CC: The next stage. If keeping things natural is a benchmark, kind of like aiming at violin acoustics in physical modeling research, then once we've got something close to the real thing it's time to have some fun with more extra-physical possibilities. In fact that's the original intent, but I'm always intrigued by what we learn along the way in the calibration-to-reality phase of the work and that's the grounding for stuff that should never happen in the "real" world. Makes sense? As composer / performer / time traveler / and audio-teleportation researcher, I can sense some new forms on the horizon. Check out the cyberSimps project by Daniel Walling for a (totally amusing) set of drama improv sketches that used the stage separation factor to create results that would never have been possible in a traditional venue.
    PT: Thanks for the CyberSimps tip. I watched some of the skits and it was pretty funny. Definitely some very interesting narrative possibilities with what they were doing. So you said above that you sense some new forms on the horizon. Wondering if you could ellaborate on that? Are there any particular forms you have in mind?

    CC: We just finished a "happening" style event during a Merce Cunningham residency here last week. It's the second one I've done involving local networks for audio (any zone connects to any zone). One was a "centralized" setup in a star and the 6 outer stages were patched through the central one. Lots of interaction throttling and mixing. The recent one was "decentralized" by contrast and this let the various "acts" pick and choose how they interacted. First form for the future would be a wide area setup for freely contributed stages and performers, in the spirit of spontaneous combinations. A second form of interest would be more composed and involve unfolding networks of real-time but determined structures, a bit like Ping but with instrumentalists.

PT: How does "Ping" differ from the collaborative performances in what it is trying to ellicit out of the network?

CC: Ping began a thread in my work with Greg Niemeyer that we've come to call Extrasensory Perceptions (there's a CD by that name with Ping and Oxygen Flute). The concept is simple: we can translate sensing data into sensory experience and make normally hidden aspects of the environment perceivable. Network flows and the carbon dioxide / oxygen exchange have been it, so far. That gives the substrate for these pieces, but it's finally about composing up from that to the surface. Since these are real-time systems, they become like interactive improvs with the data flows.

Like you said, Ping is a piece "about" the network, where SoundWIRE is more using the network (for audio connectivity) and about the network using audio (for monitoring QoS).

    PT: In making the hidden aspects of a network audible through Ping, I'm wondering how you look at the network in terms of it being an object or resource that can be tapped and tapped again for artistic purposes because of its complexity/density?

    CC: It's interesting to me for its dynamics. Whether the interest holds or I move on to other aspects of life (gas-guzzling traffic, for example) remains to be seen.

    PT: There are a lot of people out there right now working in this domain artistically, creating music, animation, interactive pieces, etc. that all don't just use the network as a medium but seem to be about the network itself. Now this is asking for a bit of time traveling (but you seem up for it :-) ), and I'm wondering what you see as the future of networks as a subject? Do you think this is just a trend because of the still relative new-ness of the technology, or is there something much deeper that the network is really reflective of (i.e., human relationships, connections, properties we see in nature, etc.) that will cause it to be an enduring subject?

    CC: The current piece has *no* network aspects at all. So I guess it's possible for me to lose it.

PT: Could you talk/write a bit about the compositional process behind "Ping" and its derivatives. Were you already working on SoundWire when the idea for "Ping" came about, or did the idea for "Ping" spawn the SoundWire project?

CC: SoundWIRE came first. It was a very utilitarian reseach project funded by NSF with the hope that we'd produce useful approaches to network monitoring. Turns out it works, though that wasn't obvious at the time and I have to hand it to the program officer for going out on a limb and supporting us. The approaches included fine-grained monitoring (with looping audio) and coarse-grained (with ping statistics mapped to tones). Once we got used to listening to the latter, it was only a matter of the right compositional project coming along before I borrowed the technique over into that side of my life. Paradoxically (?), the hundreds of thousands who have heard Ping are now aware of the network in a way that was the original intent: that there are particular instabilities and patterns that are a quality of network connections. It's just that the "research" side and the "art" side needd different support to bear fruit.

PT: What made you interested in composing for a network?

CC: The patterns. If we had perfect networks they'd sound regular and boring. The other extreme would be totally random busted networks. In between, where I find other systems to mine for music as well, is this place that's got predictability and surprise in varying balance.

    PT: I wonder if the patterns we find in networks are comparable to patterns one would find in particular natural systems?

    CC: that, plus it's density. There is no longer any chance to simulate the Internet, it's grown way too big for our current supercomputers to model. So like biological systems, we wonder about its phenomena even in fundamental terms, poking around as best we can to appreciate its behavior. Poking scientifically and artistically can be equally engaging and bring equally important kinds of understanding.

PT: I have listened to the samples of SoundWire on your website and read what I could find about it, but I'm still a bit unclear as to what is actually happening to create the sound. My notion of a typical 'ping' is a packet sent out once-a-second (by default anyway) to test network latency.

CC: ...yup

PT: With SoundWire, are you instead flooding the network with pings in order to get a more granular picture of the RTT and any latency?

CC: In a sense, but it's packets of audio instead of ping data. The audio is streaming and the "flood" in this case depends on sample rate, #chans, and bits/sample. Granularity is thus the inter-sample or inter-audio-buffer time (can be microseconds, instead of seconds). Network behavior patterns look very different on these time scales.

    PT: So then you're listening to the audio packets from some node in the loop? What is the audible effect if packets are dropped or if the rtt slows significantly? does the pitch drop or do you hear little glitches if packets drop?

    CC: exactly. You can tap it on either node / host to listen. And longer RTT = lower pitch. Small glitches from dropped packets create a timbral change like a reattack or interference on a string.

PT: How is this translated into sound? Through some translation of the millisecond RTTs into corresponding frequencies?

CC: No -- the sound comes "naturally" from sound waves created across the network. The audio stream is looping around a network path (which has a delay). It's as if the network were a real pitched instrument like a stretched string, where the pitch depends on the delay time for an excitation to loop around on the string medium. Instead it's done over the network by designing a physical model circuit (distributed over the network) and using the network delay in place of the computer memory that's typical for physical model delays. I talk about "plucking" the network, because of course it's silent sound that's recirculating until it's excited, same as a Karplus-Strong circuit that's live but not been impulsed.

PT: What is the current state of your compositional work with respect to networks and the results of SoundWire? Are you continuing to compose for the medium?

CC: One aspect has been that all this is collaborative, whether on the technical side with engineers, or on the performance side with players who I get to meet musically but not physically, or on the sound installation side working with Niemeyer and seeing it become an "instrument." The last event was a "300-square mile Recording Studio" demo'ed for AES in SF last Halloween with the music of Stanford's Mariachi players. The next is "Musicollage" March 8th, a happening with CCRMA and other musicians spread around the Cantor Arts Center in collaboration with the Merce Cunningham Dance Troupe. The tech aspect involves surround sound imparted around a loop of three stages. All "acts" will hear and be mixed with all other acts (4-hour event with dozens of acts). So it's not a composition per se, but a public event with lots of unrepeatable possibilities. Tech collaborators in this event are the folks from Network Sound, Inc. of San Jose who've developed a digital snake we'll use.

    PT: This sounds very interesting and I would definitely be there to see/hear it if we still lived there. Will it be webcast at all?

    CC: Sorry I didn't respond sooner! This was the real reason I've been delayed... I think there was an hour or so that was webcast but I can't find any recordings online at the moment.

CC: Also in the spirit is my current improv work with Roberto Morales. We're exploring layerings that we hope will involve non-physically-present collaborations resulting in "recombinant recordings" -- on the web, an improv database, growing and listenable in continuosly mutable ways.

And like the tight audio loop in the happening in March, two other concert / audio teleporation projects. The concert will be an "around the world" loop in Summer '06 with at least three stages. And the latter, a telezone booth installation for SF and Zurich, for the public to use called "Whispering Walls." It's another Chafe / Niemeyer collaboration. Amidst all that, I really do hope to write for a broken ensemble on multiple stages. The writing will be like my current project for pno,vn,vc,cl quartet --

    PT: These all sound very interesting. I understand the audio loop concert scenario, but could you explain the 'Whispering Walls' project a little more? How does the telezone booth work and what would the participants do?

    CC: WW is meant to be a place where you can go to transport yourself out of your present surroundings, acoustically find yourself "elsewhere" kind of like talking on a phone. The elsewhere is somewhere we have a hard time pinning down even if it's us talking to each other from known spots a continent apart. High-quality audio, probably ambisonic projection, possible noise cancellation all are being considered for the booths. Participants will simple walk in and listen or interact with their opposite number(s).

PT: Have you seen other network-based pieces composed using the fruits of the SoundWire project? If so, could you tell me a little about them.

CT: I mentioned CyberSimps, it's at http://www-ccrma.stanford.edu/groups/soundwire/cybersimps/

Aside from that, nothing yet that I know of.

PT: Are they mostly performance-based or collaborative works (in the vein of the cross-country performances at CCRMA) or have you also seen installation-type works closer in concept to "Ping"?

CC: Atau Tanaka has a piece called "Global String" which is a music installation but not a descendant of SoundWIRE.