Inspiration
You want to discover new music but you don't want it to affect your YouTube or Spotify recommendations. You want to avoid the hassle of changing your YouTube or Spotify settings to private. You are experiencing a mix of feelings and emotions and you want to play music that reflects that. You're in a temporary or semi-permanent headspace and you want a specific vibe.
What it does
It finds similar music you'll like based on your query (which can be a combination of artist, genre, mood, or decade) and shows them to you as a YouTube or Spotify playlist. You can listen to your new discoveries right within the app. It also keeps a history of your searches so you can go back and replay searches/generations you particularly liked.
How we built it
It was built with Vertex AI, Node.js, Express.js, Nunjucks, and Pico CSS.
Challenges we ran into
- getting output in the format I requested: I had to really polish the response because it would sometimes hallucinate and generate it's own input & output within the response or respond with the wrong format
- Ran into this a few times with safe, clean prompts: "The response was blocked because of unknown reasons. Please try rephrasing your prompt."
Accomplishments that we're proud of
- deployed a feature-complete web app https://musematch-ai-jekeihr3jq-uc.a.run.app/
- getting really good music recommendations
- creating an actual music player by integrating Spotify and YouTube
- clean and polished UI
What we learned
- Vertex API responds quickly
- The YouTube IFrame API is very robust. If you can think it, you can probably do it with the IFrame API
What's next for MuseMatch AI Music Player
- add integrations for other music services like Deezer, Tidal, and Apple Music
- let users login with Google so we can better save their history. History is currently stored locally.
Built With
- express.js
- gcp
- javascript
- node.js
- nunjucks
- spotify
- youtube

Log in or sign up for Devpost to join the conversation.