FireTV is a cutting-edge, AI-driven video streaming platform built to highly personalize and socialize the viewing experience. It goes beyond traditional streaming by utilizing AI to detect a user's current mood and instantly recommending content tailored to their emotional state. Additionally, FireTV features intelligent video summarization for long-form content and robust real-time synchronization, allowing friends to host watch parties and stream videos together in perfect sync.
Analyzes user signals (text, audio, or video) to detect emotions and suggests contextually matching videos.
Automatically generates concise text summaries and highlights key moments of long videos using Gemini.
Supports real-time, multi-user watch parties with perfectly synced play, pause, and seek controls.
WebSocket integration broadcasts chat messages, mood shifts, and playback state changes instantly to all users in a viewing room.
A blazing-fast frontend built with Vite handles the UI, video player logic, and capturing user signals for mood detection.
A Python-based FastAPI backend manages active WebSocket connections, acting as the central hub for playback events and chat messages.
The backend securely communicates with the Gemini Flash API to process video transcripts for summarization and evaluate user inputs for real-time emotional state classification.
Add WebRTC-based voice channels to the watch parties so users can talk to each other in real-time.
Introduce continuous emotion tracking via webcam that dynamically adjusts the recommended queue as the user's mood shifts.
Expand the AI summarizer to automatically inject clickable visual chapters onto the video scrubber based on topic changes.
Interested in this project?