Description:
Currently, Streamify shows random recommendations based only on the artist of the currently playing track.
We want to upgrade this by adding a smart AI recommendation system that suggests songs users are more likely to enjoy — based on their listening history, liked songs, and track similarity.
Goal:
Enhance the recommendation engine using a lightweight ML model or API that analyzes user behavior and song metadata (e.g., genre, mood, tempo).
Proposed Solution:
-
Collect basic listening data (recently played, liked songs, skipped songs).
-
Use this data to calculate similarity scores between tracks using:
Cosine similarity on embeddings (artist, genre, mood, etc.), or
An API such as Spotify Recommendations or JioSaavn similarity endpoint (if available).
- Display “Recommended for You” or “Because You Liked…” section on the homepage or player screen.
Tech Suggestions:
Use a small ML model (like k-NN) for local recommendations.
For scalability, prepare backend endpoint /api/recommend that returns smart suggestions.
Integrate AI/ML libraries (e.g., scikit-learn, or sentence-transformers if embeddings are available).
Description:
Currently, Streamify shows random recommendations based only on the artist of the currently playing track.
We want to upgrade this by adding a smart AI recommendation system that suggests songs users are more likely to enjoy — based on their listening history, liked songs, and track similarity.
Goal:
Enhance the recommendation engine using a lightweight ML model or API that analyzes user behavior and song metadata (e.g., genre, mood, tempo).
Proposed Solution:
Collect basic listening data (recently played, liked songs, skipped songs).
Use this data to calculate similarity scores between tracks using:
Cosine similarity on embeddings (artist, genre, mood, etc.), or
An API such as Spotify Recommendations or JioSaavn similarity endpoint (if available).
Tech Suggestions:
Use a small ML model (like k-NN) for local recommendations.
For scalability, prepare backend endpoint /api/recommend that returns smart suggestions.
Integrate AI/ML libraries (e.g., scikit-learn, or sentence-transformers if embeddings are available).