Thinking Tech

A distress detector for emergency call centers

A distress detector for emergency call centers

Posting in Design

Using software voice analysis, a new system for emergency call centers can help zero in on the most urgent calls.

From the perspective of the person placing it, a typical 9-1-1 call goes something like this: the phone is answered, standard questions are asked, details are gathered and emotions are soothed. Assistance is sent. The caller has a singular goal--to get help--and expects, reasonably, that the dispatcher shares his interests.

Of course he does. But from the dispatcher's perspective, the situation is much more complicated. He must figure out what kind of response, if any, is appropriate; he has to note location data provided by his call center's equipment; and he has to somehow decide how to prioritize the call, particularly in large-scale emergency situations where resources strained.

Made by a human, these judgments are uncomfortably subjective. And in situations like natural disasters or terrorist attacks, when emergency services can be severely overloaded, the perceived urgency of a call might mean the difference between a situation being dealt with or ignored, at the peril of its victims.

A team of researchers at Delft University might have found a way to help. They've developed software they claim can measure the urgency of a call--specifically, the stress of the caller--to give call centers another tool for dealing with sudden surges in contact. Say the researchers:

Stress and negative emotions, in general, have a strong influence on voice characteristics. Because speech is a natural means of communication, we can utilise the sound patterns of speech to detect stress and (negative) emotions in a non-intrusive way by monitoring the communication

Factors like speed, breathing rate and pitch changes are all correlated with stress, something which humans intuitively understand. Our judgement of these speech properties, however, is crude and unreliable. By analyzing and learning from large databases of recorded emergency calls for which the outcomes are known, their software is able to determine the seriousness of the call with a surprising degree of accuracy--according to the researchers' published study, its equal error rate is just 4.2%.

If the idea of a computer arbitrating something as personal as panic or urgency makes you uncomfortable, consider that the system is designed as a supplemental tool for emergency call centers, meant to aid the people taking calls in doing something that they're already accustomed to doing: prioritizing calls during periods when the demand for help and attention outstrips available resources. In a dispatcher's view, more data can't hurt, especially in the throes of an emergency.

And anyway, according to the researchers the technology's first applications will likely be in the military, so it'll have plenty of time for refining before it finds its way to your local police station.

Photo by Flickr user roland

Share this

John Herrman

Contributing Editor

Contributing Editor John Herrman is a freelance writer based in New York City. He is also contributing editor at Gizmodo. He holds a degree from the University of Edinburgh. Follow him on Twitter. Disclosure