Project Overview
This project exposes a machine learning model behind a FastAPI interface to predict the genre of uploaded audio files. It provides health, readiness, model metadata, and audio-inference endpoints for practical deployment workflows.
The pipeline includes model and encoder loading, feature extraction from WAV inputs, and inference orchestration in a structured API architecture designed for iterative model updates.
It's not yet finished, it lacks more robust error handling and a better model to work on more specific cases, like more niche genres.
Tech Stack
Python
FastAPI
Uvicorn
XGBoost
scikit-learn
Librosa
Pandas