Tempr

AI-powered music app that generates personalized queues from your mood, photos, and calendar

Project Overview

Expo (React Native) · TypeScript · Gemini API · Spotify Web API · Supabase Auth

Tempr solves Spotify Choice Paralysis — that feeling of endlessly scrolling your liked songs and finding nothing that fits the vibe. Users describe a mood, upload a photo, or let the app read their calendar, and Tempr uses Gemini to convert that context into a target vibe vector. It then runs a cosine-similarity nearest-neighbor search across the user's Spotify library (using audio features like energy, danceability, valence, and acousticness) to queue the best-matching tracks directly in Spotify.

Demo Video

Live Demo

View the live Tempr web interface.

Open Live Site

GitHub Repository

View on GitHub

Key Features & Impact

Mood-to-queue generation via text prompt or photo using Gemini's vision and language models
Calendar integration that auto-generates playlists for upcoming events with a 10-minute notification
Cosine-similarity vector search over Spotify audio features for accurate vibe matching
Memories library to save and re-queue past generated playlists in one tap
Discover mode with swipe-to-like and YouTube video previews for finding new music
Batched Spotify searches and AsyncStorage caching to keep generation fast and reliable