Spotify is working on new algorithms capable of recommending music depending on our emotional state. The latest Spotify patent is based on a voice analyzer, which would allow data such as emotional state to be obtained to suggest new songs to listen to.
The BBC has revealed that in 2018 Spotify registered a patent to generate recommendations based on “observations” of the user, and was approved on January 12, 2021, and first spotted by Music Business Worldwide.
According to Spotify, the methods currently used are “unsatisfactory”, because they still require the user himself to configure the recommendations by entering data such as what his favorite groups are, how old he is, what is his gender, or what is his mood.
The solution, according to Spotify engineers, may lie in the use of speech recognition to obtain these and other data about the user, based on “contextual clues” that are not always obvious.
Based on the intonation, the stress detected, and the rhythm in which we speak, Spotify believes that it would be possible to obtain a series of metadata, including various categories of emotions, gender, age, and accent; but those are just examples, and the patent lets it be understood that there would be other characterizations that would be helpful.
These data would be combined with other more ‘basic’, such as the songs that we have already played, or even the personal tastes of our friends on the platform, to improve the recommendations.
At Spotify, they believe that the music we prefer for each moment depends a lot on the type of person we are and how we feel. For example, their research would have found that introverts tend to focus on exploring an artist’s catalog, listening to more songs by each artist they discover. Based on that, the lists generated for this type of person can include several songs by a single artist.
For the Spotify app, it would be easy to ask permission to use the microphone and obtain information that would allow you to create custom lists. However, the company itself seems to be aware of the privacy concerns it can bring with it.
In the patent, it recognizes that a user’s “digital history” is something “extraordinarily personal” and that it should be treated with consideration for possible misuse. Therefore, the company’s researchers reject future studies that may violate the ethical standards of data use and privacy.
