Want to feel upbeat? How emotion-aware algorithms can help

University of Melbourne, Faculty of Engineering and Information Technology

New research from the University of Melbourne has found that playlists curated by algorithms can help people regulate their emotions up to 69 per cent of the time. With emotional distress on the rise, the research holds promise for using algorithms to improve emotional well-being.

With music streaming apps such as Spotify being used by millions of people every day, algorithms are increasingly shaping how people listen to music. Dr Xanthe Lowe-Brown, who recently completed her PhD in the School of Computing and Information Systems, investigated whether algorithms informed by psychology and ethical principles could guide users towards desired mood states.

She found that curated playlists were significantly more effective at helping users reach their desired state than self-ordered playlists.

"Music streaming platforms provide an almost infinite selection of music available at the click of a button. We want to find a way to harness this power to help brighten listeners' mood," Dr Lowe-Brown said.

She and her collaborators built a pilot app, MoodDJ, for a field trial, which worked as a web app connecting to volunteer trial participants’ Spotify.

Users would input their current and their desired mood using two factors – their ‘arousal’ (how much energy they had), and their ‘valence’ (how positively or negatively they felt).

MoodDJ then drew on users' music libraries to curate playlists using Spotify’s data on a song’s energy and valence, aiming to bring users from their current state to their desired state.

The app showed short-term benefits, with many users appreciating how the app prompted them to reflect on their emotional state when using it.

“Our initial findings highlighted that while people reported feelingthat the benefits of an app like MoodDJ outweighed the risks, it’s crucial that this kind of technology considers privacy and transparency concerns,” said Dr Lowe-Brown.

“Algorithms in Spotify and Apple Music are increasingly used to curate mood-based playlists, but they need to be informed by psychology if they want to have the impact they claim to have.

“Equally, it’s important to consider questions regarding how user data on emotions-based listening is being stored and used, to be sure vulnerabilities aren’t being exploited.”

Now, with $15,000 in funding through the Faculty of Engineering and Information Technology’s PhD Lab to Launch Program, Dr Lowe-Brown is using findings from the pilot app to develop MoodDJ into an app for public use, opening doors to a range of applications in both personal listening sessions such as to energise oneself on the morning commute and in public spaces such as waiting rooms and gyms.

Read more here.

Lily Petrovich (Communications Officer) | [email protected] | +61 405 406 900