Cybersleuths Rediscover Information-laden Audio

Throughout the history of the animal kingdom, sounds have always been full of meaning. They provide valuable information on the whereabouts of prey and predator, the location of family, the intentions of friend and foe. Sound cues have always been as important as information obtained by scent and sight.

Until the computer revolution, that is. Computers and their designers have mostly ignored sound's potential for conveying meaning. Until soundcards became widespread, the only sounds the typical computer could make were a few irritating chirps and beeps. Even after soundcards became fashionable, their sonic offerings were mostly limited to the cheesy sound effects of computer games.

Now researchers at IBM's Computer Music Center in Yorktown, New York have rediscovered the eternal truism that sound is full of meaning. They are reworking Sonnet, a new version of an old visual programming language that allows the addition of sound to computer programs, which are overly rich in visual information. The cluttered appearance of most desk-top computers' operating systems is a symptom of excessive visual stimuli.

IBM scientists are attempting to use sound cues to convey information without adding further clutter to the screen. The process is called "sonification." Pointing out that we respond automatically to sound cues even while performing other tasks, project manager David Jameson says: "If you're driving . . . you change gears when the engine revs too high, without thinking, even while talking to a passenger."

The scientists believe we may be near overload in our ability to assimilate visual images. They would like to convey simple information such as rising or falling temperatures or stock prices with aural tones instead. Sonification could be the "glue" that connects and integrates diverse parts of programs, the way narration, dialog, and music connect otherwise disparate visual images in a film.

Not all computer experts agree that sonification has much potential. It's too memory-intensive, say some, like International Data Corporation's Richard Kay. "Most computing being done today is character-oriented. Since characters are lightweight, they can be compressed and compiled," Kay said. "But when you start bringing in sounds, you are committing to systems that accommodate heavyweight data in all ways." An hour of digitally stored music needs about 650 megabytes of memory.

Most of the research at the IBM lab centers around what sorts of sounds are most appropriate for conveying particular types of information. Eric Scheirer, a research assistant at MIT's Media Lab, noted that although today's computers are relatively primitive in the sound department, colleagues like Jameson are working on the next generation. "There's a good chance they will find a market for Sonnet," he said. "It could absolutely further the field of sonification."