What if, instead of tapping at different songs and being at the mercy of playlist recommendations, Spotify allowed users to alter their recommendations from the UI itself?
This project is a proof of concept that demonstrates how users could potentially edit their song recommendations based on certain parameters.
At times, a user might feel like listening to songs that are:
- More energetic or fast-paced than Spotify's current recommendations
- More instrumental than usual
This demo allows users to adjust two key parameters:
- Energy
- Instrumentalness
- The app sets initial values for 'energy' and 'instrumentalness' based on the user's existing listening history.
- It then seeds existing recommendations with the new values chosen by the user to suggest new songs.
Formula: Existing recommendations + User-chosen values of Energy & Instrumentalness = New recommendations
- User Control: Highly discerning users (often premium subscribers) would appreciate having some control over what they listen to in a particular session.
- Reduced Churn: Giving users more control could potentially reduce churn rates.
- Transparency: As people become more aware of how algorithms determine their habits, there might be a shift towards allowing users to fine-tune their recommendations.
With the advancement of Large Language Models (LLMs), users might even be able to fine-tune their recommendations using natural language in the future.
We're happy to hear your views and comments on this concept. Feel free to open an issue or submit a pull request!
This is a proof of concept and is not affiliated with or endorsed by Spotify.