Voice-Activated Mood-Based Playlist Creation for Spotify
Voice to Playlist, a mic on the Spotify home screen that turns a spoken mood into a playlist
✳The problem they went after
Getting music that matches an emotional state means manual search, generic mood playlists, or browsing until something sticks. That is hardest for mood-driven listeners, commuters, and gym-goers, who have neither the time nor the vocabulary to turn a feeling into keywords. The gap shows up as frustration, cognitive overload, and people closing the app when the music does not fit the moment.
✳What they actually did
The design starts with a microphone icon on the home screen. A user speaks naturally, asking for a long-drive playlist or upbeat workout songs, and the app detects intent, analyses mood, and generates a curated playlist with options to play, refine by voice, or save as a favourite. Skips and likes feed back into later recommendations. Rebecca maps four segments the feature serves: mood-driven listeners, hands-free users like drivers and gym-goers, casual listeners who want music without curating, and discovery-focused users open to new artists.
✳What you can take from it
Worth stealing
- Full voice workflow specified from mic tap to saved playlist
- Refine-by-voice loop instead of one-shot generation
- Skips and likes feed later personalisation
- Four user segments defined for one feature
Read this if
PMs writing an AI feature spec that needs a concrete end-to-end flow
The transferable bitEvery step is named, from the mic tap through intent detection to the refine-by-voice loop and the feedback signal that improves the next result, which is what specifying an AI feature as a workflow rather than a capability looks like. The business case stays plain too: daily active users, session length, and Premium conversion, with the feature applicable across languages.
✳The submission itself
Open the PDF →Reading on a phone? The embedded viewer is cramped — open the PDF in a new tab instead.
