Blue Monday?

Spotify wants to use your voice to suggest content based on your mood

Do we want Spotify to analyze our speech to decide how we're feeling?

Shutterstock

According to Music Business Worldwide, a newly granted Spotify patent outlines how the music-streaming company is seeking to develop technology that would be able to assess the "emotional state, gender, age, or accent" of Spotify users, and then make recommendations about what they might want to listen to based on their current mood and historical taste.

The patent was filed in 2018 and the proposed feature it seeks to protect is described in the supporting documents as a "method for processing a provided audio signal that includes speech content and background noise" which is followed by "identifying playable content, based on the processed audio signal content." The patent also adds that the data collected and analyzed includes "intonation, stress, rhythm and the likes of units of speech."

Amazon got there first, sort of — You've may have heard of another example of this sort of personality or mood analysis tech that already exists: Amazon's unnecessary and overwrought Halo fitness tracker. In addition to counting steps and the other basic you'd expect from such a device, the wearable attempts to assess and record your mood by analyzing the tone of the wearer's voice at random points throughout the day. It even offers (sometimes) patronizing suggestions based on that information, say for instance that the wearer might want to try and "sound happier" if it deems them to sound too dour. It's hyper-personal, with a healthy dose of creepy and invasive.

Your data and metadata — Spotify's patent not only takes account of a user's voice but also the background noise. Sounds like traffic, birds, office noises, or other environmental cues will be taken into consideration for the content suggestions that follow the patent shows. It even notes if the user is alone or situated in a group setting, like, say, at a party. You can see where this is going. If you're at work, you'll get different suggestions than you will if it seems you're at a picnic with friends.

The data will be used not only to make song suggestions but also to refine audio advertising content. You know, those pesky commercials that interrupt your playlists on Spotify's free tier, or on some of its original podcasts. Ads targeting Spotify users could become the audio equivalent of Instagram ads with regards to their specificity.

This shouldn't really come as a surprise. Spotify is in the business of making money, it does that via ads and by encouraging paid subscriptions thanks to attractive content. This sort of customization was inevitable.

We're not so sure about this — Even beyond unnervingly personalized music suggestions, it's hard to imagine an algorithm accurately assessing a person's accent, tone, and mood through recordings. But then, it was hard to imagine commercial flight a century ago. More worrying, really, is how streaming services seem to be reducing the odds of the accidental discovery of new content, and the joy such accidents can bring.