TextFM with Graham Harwood

Though you might not be aware of it if you live in the U.S.—where mobile-phone technology is still a creaky Tower of Babel—“texting” is a massively popular phenomenon in the rest of the industrialized world, especially among young people. Formally known as “SMS” (for Short Message Service), texting is a way to send text messages from one mobile phone to another quickly, easily, and cheaply. There are currently more than 30 million text messages a month being sent worldwide, and that number is expected to rise to more than 100 million in the next two years.

TextFM, a “simple, lightweight, open media system” designed by Londoners Graham Harwood and Matthew Fuller, takes advantage of the widespread availability of SMS-capable mobile phones to allow people to broadcast “voice” messages over the public radio airwaves. Using TextFM is simple: you send a normal text message to a central phone number, where it is captured by a computer. The computer converts the message to speech using voice-synthesis software, and your spoken text is then sent to a transmitter and broadcast over an FM radio frequency. As part of your message you can also include several optional codes (“switches”) specifying the language your message is in, which of ten voices to use, the pitch of the voice, and the speed at which you want the text read.

The TextFM software is non-proprietary, “open source” code, meaning it can be freely downloaded—and even customized, if necessary—by anyone. Anyone with access to a computer running the Linux operating system (which is itself free, open-source software) can set up their own TextFM “server.” Installations are currently running in Vienna, London, and Amsterdam, with more locations in the works. One of the current goals of the project is to grow a decentralized network of TextFM servers around the world. After a message is received and broadcast at one TextFM site, it can then be forwarded to other sites in the network for broadcast there.

I spoke to Graham Harwood (who is currently doing a residency at De Waag in Amsterdam) during a recent visit to New York, where he gave a presentation on TextFM at the Museum of Modern Art.

Dave Mandl (Rail): Can you describe how TextFM servers in different locations would work together?

Graham Harwood: The server doing the voice synthesis sits there [in Amsterdam], and so people text to my phone, my computer reads the text messages straight off, then sends those streams to the server in Austria [where they] join the stream of people texting there. And the same is happening there, on their server. So it’s looking more and more likely that you can have different nodes for this device. Because one of the big problems has been getting around the airwaves problem [i.e., getting access to radio frequencies to broadcast over]; the radio thing is a complete nightmare.

Rail: That’s interesting, because one of the original goals of the project was opening up the airwaves. So do you now see the future being more in webcasting these messages, streaming over the net rather then continuing with the radio model?

Harwood: No. Generally it’s a localized project. [Local administrators can send messages] off into radio, or off into a public announcement speaker system, or some other viable way for the local area. Because the laws on radio are so very different between different borders and different places, there’s not a kind of one-solution-fits-all. It looks like you’ve got to have a lot of different elements of the project that can be locked together in different ways to suit local environments. It could be in a public address system in a particular environment, it can be in a club, you can use a CB.

Rail: So it’s completely decentralized and autonomous: “Here’s your stream; do what you want with it. If you have access to some radio frequency, then broadcast it. If you want to webcast it, do that.” What kinds of messages have people been experimenting with?

Harwood: One kind of speculative notion would be if we can set up a series of speakers aimed at a public building here, or a public monument or something, and do the same in a number of countries, and then use these different nodes to actually just send shit to these public address systems, it would be a really good method of—

Rail: An audio bulletin board.

Harwood: Yeah. Because a lot of people in Vienna use texting as they’re walking past the public-address system there to just write in their text message that just booms out in that locality. So it’s almost like grafitti-ing as you walk past. And one of the really invigorating notions about SMS is that everyone has their own remote in their pocket, you know, as you walk past some kind of bulletin board, some kind of address system to just leave something, post something, place it there, in a mobile space. And that is really a kind of social dynamic, because it gets it back out in the streets, out of your bedroom, and your screen.

What’s interesting about it is the complete system, it’s not the content of the system. It’s the media systems that are being brought into play for particular purposes. And the content of it is kind of secondary. For me if it’s particularly geared at the physical object or a physical space, then I’ll quite happily send a steam of Bush probability speaking [a Harwood project that creates ersatz Bush speeches based on word frequencies in previous Bush speeches], or some other activity. And so I think they’re the really core interests for me, and it also came about because of this thing of wanting at first to create a local media system, and then seeing how people wanted to actually interact or manipulate that system. Not just content. And that became part of the project.

Rail: What do you mean by “manipulate the system”?

Harwood: I mean being about to change voice, trigger events, change pitch of voice. We did one experiment with a group of students where we took this trip of Bush's to come to a South American country and combined [his speech] with a bunch of other robots crawling other websites, and put that [material] together—

Rail: So you just inject it into the stream?

Harwood: Inject it into the stream, yeah. At timed intervals. And of course you get these kinds of reactions to it from people texting. So it’s not a completely “open” system, but it’s a system that’s using language as data, and then allowing people to interrupt that.

Rail: You’re going to be doing something with Resonance FM [a new community radio station in London]?

Harwood: Yeah, we’re going to do it with Resonance. I think we’re going to use nighttimes.

Rail:  You mean in a time slot between the hours of so-and-so?

Harwood:  In the different kinds of testing we’ve done, we’ve seen that TextFM works really badly in some environments and really well in others. That’s quite interesting in itself. If you only have a three-hour time slot somewhere and you just do it, it’s crap. Because the network doesn’t develop. If you do it, though, in a kind of closed conference-type session, it works very well. Like where there is a particular subject and you use a local PA system, and people are dropping their messages into it. It works really well like that. Where it works the best is when you’ve got something ongoing over a month period or something like that, where it can build up its own clientele. If you’ve got a specific action with a public-address system against a particular building, that works very well. But these light encounters with it in public spaces are bad. Because people don’t get it.

Rail: This project seems more humanist, in a way, than the net, just because there’s a voice involved—though I haven’t heard it; I don’t know how synthesized and cyber-punk it sounds.

Harwood: The aesthetic of voice synthesis is bad. A lot of people hate it. I went through a thing of really hating it, but then I began to like it because it’s like the country-and-western of the cyber world. It’s naff, it’s tasteless, and it grates. That’s one of the things in Amsterdam—I’ve done it at some reasonably bourgeois events. And people kept turning it off, because they found it so annoying, and I was in heaven. And people get really scared of it as well, because once you alter the pitch and rate of the thing, you get into some really grating, tasteless aesthetics, which I have a fascination for, social elites’ use of aesthetics. Also I did things like use a lot of harmonies with the voice synthesis, with jingles and stuff. So those horrible synthesized voices are actually singing harmony with a TextFM jingle. And we use birdsong, British birdsong, as the studio track. So that’s the background all the time in TextFM. Because birds kind of have these intricate media systems by which they declare territory and intention. It’s also like the music sound of the twittering of the bids. So it fits really well. “What kind of aesthetic can you choose for such a system?” And birdsong seemed to be the most stupid and appropriate. [Laughs.]


Dave Mandl

DAVE MANDL was the Rail's former music editor. He is a freelance writer/journalist.