Skip to main content

New algorithm could help diagnose depression by analyzing the tone of your voice

hotline dating app talking on phone 123rf 45715075 ml
sifotography / 123RF
The concept of an AI psychoanalyst has been in circulation for decades, tracing all the way back to Joseph Weizenbaum’s ELIZA chatterbot in the 1970s. But now researchers from the University of Southern California are taking the idea to the next level, courtesy of a machine learning algorithm designed to analyze a person’s speech patterns and help diagnose the possibility of depression in the process.

The tool is part of an ongoing research project called SimSensei, referring to a Kinect-powered virtual therapist able to “read” patient’s’ body language for signs of anxiety, nervousness, contemplation and other emotional attributes.

More recently, however, the project has increasingly focused on not just understanding the responses given (like Apple’s Siri does, for instance), but also the manner in which they are spoken. “I’m not so interested in what people say, as how they say it,” Stefan Scherer, one of the researchers involved with the work, tells Digital Trends. “We’re focusing on aspects of speech like voice quality — from the timbre to the color of the voice: whether it’s a tense voice, a harsh voice, or a breathy voice. We want to pick up these changes and contextualize them.”

Scherer calls his work “behavioral analytics” and says that it’s all part of creating a more fully-realized tool which can be used to augment the abilities of a real therapist or physician. “It provides a different set of eyes and ears that they would not normally have available,” he says.

In a recent paper, the authors of the study explain how: “depressed patients often display flattened or negative affect, reduced speech variability and monotonicity in loudness and pitch, reduced speech, reduced articulation rate, increased pause duration, and varied switching pause duration. Further, depressed speech was found to show increased tension in the vocal tract and the vocal folds.” Such vocal tics may not immediately be picked up on by a human.

Looking forward, Scherer says he could see technology such as this being installed in smartphone apps, so that people can more objectively measure moods in a similar way to how the “Quantified Self” movement currently does health-tracking. “You could imagine people asking if they’ve done their 1,000 smiles in a day, or whether or not they are getting excited about things,” he says. “It could be used for both people suffering from depression but also for the general population.”

Editors' Recommendations

Luke Dormehl
I'm a UK-based tech writer covering Cool Tech at Digital Trends. I've also written for Fast Company, Wired, the Guardian…
Man vs. machine: An A.I. algorithm attempts to break a world speed record
algorithm designing bike speed record gettyimages 173175649

Artificial intelligence helps design an ultra-aerodynamic bike

Artificial intelligence algorithms may be able to beat grandmasters at chess, defeat champion players at Jeopardy!, and vanquish the world’s best Go players. One thing they can’t do: Beat the world’s zippiest cyclist for the speed record of the fastest human-powered vehicle at 83.13 mph. After all, the record includes the word “human” in its name.

Read more
How A.I. and a prehistoric creature could help predict animal behavior
ai animal behavior hydra copy

Like the sentient toys in Toy Story, animals seem to lead exciting lives when humans aren't watching. Ancient creatures called Hydra, on the other hand, don't seem to do much at all. But that's what makes them so interesting.

In a recent study, researchers used a machine learning algorithm -- one that is usually used to filter spam email — to catalog Hydra behavior. With one of the simplest nervous system on Earth, Hydra give scientists a basic sample to study how neural activity relates to physical activity, potentially paving the way for future research into predicting animal behavior.

Read more
MIT’s new A.I. could help map the roads Google hasn’t gotten to yet
mit mapping roadways tech boston with roadtracer

RoadTracer: Better Automated Maps

Google Maps is a triumph of artificial intelligence in action, with the ability to guide us from one place to another using some impressive machine learning technology. But while the routing part of Google Maps doesn’t need too many humans in the mix, manually tracing the roads on the aerial images to make them machine usable is incredibly time-consuming and mundane. As a result, even with thousands of hours spent on this task, Google employees still haven’t managed to map the majority of the 20 million-plus miles of roadways that stretch around the world.

Read more