StressSense is a system developed by Hong Lu of Intel and colleagues that is programmed to recognize stressed voices over the phone.
The program is able to decipher between a stressed and a calm voice by first having a user record three minutes of relaxed conversation or casual reading. The system, which comes with preprogrammed information on the vocal changes brought on by stress, can compare later conversations with the recorded reading to determine the level of stress in a voice.
Research has shown that human vocal production is influenced by stress. So placing microphones in phones provides an opportunity to continuously and noninvasively monitor stress in real-life situations.
The prototype was tested out on volunteers who participated in a mock job interview meant to induce stress. Then they were assigned a job recruiting people outdoors.
Researchers found that the prototype was able to recognize stress in a person’s voice with an accuracy of 81 percent indoors and 76 percent outdoors, where sound quality is poor.
New Scientist reports:
“The team plan to make the system a plug-in to an Android app called BeWell, which uses a phone's accelerometers and GPS sensors to record people's activity and sleep levels. People will be able to set StressSense to either listen to their voice throughout the day, or to only activate when they are having a phone conversation.”
The software can be used as a way to help people become aware of and manage their stress.
StressSense will be presented at the Ubicomp conference in September in Pittsburgh, Pennsylvania.
Via New Scientist
Photo via flickr/Ollie Crafoord