Moodify – AI-Powered Emotion-Based Music Recommendation System -- 2
Budget: $50 – $80 AUD
I’m developing “Moodify,” a full-stack web application that listens to facial cues through the webcam, recognises the user’s emotion in real time, and then queues music that fits the mood. The core tech stack is already defined—React.js (bootstrapped with Vite) on the front end, Node.js with Express.js on the back end, MongoDB for persistence, and Face-API.js together with TensorFlow.js for the emotion model.
Some details you should know before diving in:
• Authentication must rely on JSON Web Tokens; sessions, cookies, or OAuth are out of scope for this build.
• Music playback and catalogue data will come from YouTube Music; any helper libraries you favour are fine as long as they respect YouTube terms.
• Users can still override the detected emotion manually, but the UI for that should be simple radio buttons rather than a dropdown or mood wheel.
• The buffering logic that smooths out rapid shifts in expression is already sketched in pseudocode; I’ll share it, and I’d like it hardened and unit-tested.
Deliverables I need from you:
1. Fully working front-end and back-end repositories (separate or monorepo, your choice) with clean, readable code.
2. Integrated YouTube Music playback that automatically switches tracks when the detected or selected mood changes.
3. Secure JWT sign-up / sign-in flow, including token refresh.
4. Persistence layer covering favourite tracks, mood history, and user profile.
5. Clear README with instructions for local setup, environment variables, and deployment to a typical Linux VPS.
6. Brief test suite (Jest or similar) demonstrating that emotion buffering and auth flows behave as expected.
If you’re comfortable with React, Node, computer-vision libraries in the browser, and you have handled third-party music APIs before, this should be a fun build. Let me know how you’d structure the project and roughly how long the full implementation would take you.
Some details you should know before diving in:
• Authentication must rely on JSON Web Tokens; sessions, cookies, or OAuth are out of scope for this build.
• Music playback and catalogue data will come from YouTube Music; any helper libraries you favour are fine as long as they respect YouTube terms.
• Users can still override the detected emotion manually, but the UI for that should be simple radio buttons rather than a dropdown or mood wheel.
• The buffering logic that smooths out rapid shifts in expression is already sketched in pseudocode; I’ll share it, and I’d like it hardened and unit-tested.
Deliverables I need from you:
1. Fully working front-end and back-end repositories (separate or monorepo, your choice) with clean, readable code.
2. Integrated YouTube Music playback that automatically switches tracks when the detected or selected mood changes.
3. Secure JWT sign-up / sign-in flow, including token refresh.
4. Persistence layer covering favourite tracks, mood history, and user profile.
5. Clear README with instructions for local setup, environment variables, and deployment to a typical Linux VPS.
6. Brief test suite (Jest or similar) demonstrating that emotion buffering and auth flows behave as expected.
If you’re comfortable with React, Node, computer-vision libraries in the browser, and you have handled third-party music APIs before, this should be a fun build. Let me know how you’d structure the project and roughly how long the full implementation would take you.