herzify model
Music picked by mood, not by genre
The AI behind Herzify. It learns how you feel and how you listen, then picks what you should hear next: mood in, playlist out.
- role
- Design & development
- type
- Recommendation model
- models
- GPT · local Ollama
- stack
- Next.js · LLMs · Spotify API · Firebase
- year
- 2026
01 /
How it works
The model turns a feeling into a playlist, and gets better every time you listen.
- You say how you feel, or it reads it from your listening
- The mood is matched against your listening patterns
- The model picks tracks for that mood
- Spotify plays them
- What you skip and replay feeds back into the next pick
02 /
Inside the model
Mood → playlist engine
Matches music to how someone feels, not just to a genre.
Listening-pattern learning
Learns from what you actually play, skip and replay.
GPT + Ollama
Runs through an API or on a local Ollama model on your own machine.
Spotify integration
Recommendations land straight in a playable Spotify playlist.
03 /
AI know-how it proves
- Mood-based recommendation
- Learning from behaviour
- LLM reasoning & prompting
- Local models with Ollama
04 /
Next
MIDI playlists: the same mood model, picking sounds and patterns for producers instead of songs.