The Xpression app phone display, with the bottom box detailing the emotion analyzed as present in the speech sample.

Story highlights

New smartphone app allows users to record their mood through their speech acoustics

The app is designed to be used in the clinical treatment of stress and depression

It could replace manually-recorded "mood diaries", which patients often leave incomplete

App may also appeal to consumers interested in self-tracking technology

A British firm is developing an app that will allow smartphones to track their owners’ moods through their speech patterns, in the hope it can improve aspects of mental health treatment.

The app, known as Xpression, will record the user’s voice throughout the day, whether they are making a phone call or not, analyzing the acoustics for emotional content.

Matt Dobson, co-founder of EI Technologies, said the app was designed to provide a more objective, reliable record of patient’s emotions than “mood diaries” currently used in the treatment of people with anxiety, depression or stress.

“Over half of adults in the UK suffer from severe stress once or twice a week so there’s a lot of opportunity in the market,” he said.

“I wanted to know: could we bring something into the market that would help people sort out their stress and depression and anxiety? … I found that voice-based recognition was something that would work on a smartphone.”

Psychologists often ask patients to keep track of changes in their emotional state by noting them down in mood diaries, he said, but medical literature suggests only 30-50% of patients provided enough information in their diaries to be useful in treatment.

“This replaces that old technology. This is the turbo-charged version of manually recording your emotions,” he said.

Tracey Parsons, a clinical psychologist working as clinical adviser on the project, said the major problem with mood diaries was that people simply don’t fill them in.

“They don’t believe they’ll be useful and it’s too much of a hassle, or they’re so unable to recognize their own feelings that they struggle to complete them,” she said.

The app, which would activate when it recognized the user’s voice, would provide the patient and psychologist a more complete record of changes in emotion and mood, allowing the patient to get a better sense of their emotional triggers.

“It’s useful to have that to allow you to think: ‘How did I react? What was I thinking? Why did I get so angry about it?’” says Parsons.

“This would allow for more effective use of mental health services, making the app a potentially attractive proposition for health systems around the world,” she said.

Read also: Harvesting rubber from dandelions

The app does not record the speaker’s words, said Dobson, but rather monitors the acoustic elements of their speech – such as volume, intensity, pace and pitch – to take an emotional reading.

“We look for acoustic features within your words. We don’t know what you’re saying,” he said.

Snapshots of the user’s acoustic patterns were sent to a remote server, where the user’s emotional state would be calculated by a machine-learning system.

Professor Stephen Cox, head of the speech processing laboratory at the University of East Anglia and scientific adviser to the project, said the technology was made possible due to advances in machine-learning algorithms.

“This is an area called paralinguistics where one is extracting things from the voice which are not said verbally but convey meaning – one being emotion,” he said.

“All we need to do is collect lots of angry speech or happy speech. We show them to the classifier and the algorithm sorts it out.”

Dobson said one of the biggest unknowns facing the project was how willing patients would be to expose themselves to constant monitoring through their phones – even though only data relating to the acoustics, rather than the words themselves, would be recorded.

“What we don’t know is: ‘How intrusive is that? What’s the trust required to do this?’”

However, he saw the app as belonging to a group of existing smartphone products that monitored the human body, such as jogging or sleep monitors, which had gained widespread acceptance. The apps were a product of the “Quantified Self” movement, which saw users employ self-tracking technologies to gain a better sense of how their bodies responded to different inputs.

He said if a consumer version of the app was released, targeting those interested in self-tracking, it could hit the market in the second half of the year. An app suitable for use by mental health services would take a little longer due to testing requirements, he said.