Inspiration
There were two main inspirations for this project. Many people suggest that AI tools like ChatGPT are very good at suggesting new music to listen to based on past experiences and genres. There was also another similar project that was ideated in Cal Hacks 9.0 which does music recommendation based on heartbeat rhythms using Zepp OS's watch sensors. This time we decided to do a similar song recommendation system but based on a Journal entry.
What it does
Recommend songs based on your mood as it is presented in your journal entries.
Tech Stack
- Frontend: React, Tailwind, Auth0, SpotifyAPI
- Backend: Convex, FastAPI, Javascript, Python, OpenAI API, Genius API
- Database: CockroachDB, Milvus
Challenges we ran into
Getting together.ai to properly clean our lyric texts. Creating a Milvus database without an index. Getting the Spotify API to work properly together with authentication and Convex.
Accomplishments that we're proud of
Finishing our app and having a working product that we ourselves would actually use.
What we learned
How to generate and store vector embeddings in a vector database; how to integrate different APIs, such as the Spotify API, into Convex.
What's next for Rythm Moodify
- Integrating a user's playlist for personalized recommendation.
- Ability to pick and choose generated choices.
- Integrating real-time recording feedback
Log in or sign up for Devpost to join the conversation.