All projects
AI / Data·2024
Sentivox, Speech Emotion Detection
Real-time ML web app that detects a speaker's emotion from their voice.
Overview
A full-stack application that records or uploads audio, extracts acoustic features and classifies the speaker's emotion using machine learning in near real-time.
Built on FastAPI with a Librosa + scikit-learn pipeline, it includes user/admin roles, JWT auth and auto-generated API docs, genuine applied-AI wrapped in a real product.
What I built
- Audio feature extraction and emotion classification (Librosa + scikit-learn).
- FastAPI backend with JWT auth and auto-generated OpenAPI docs.
- Record/upload web front end (Tailwind + vanilla JS) and admin roles.