Overview
Spleeter is a Deezer-developed source separation library, built in Python using TensorFlow. It provides pre-trained models for separating audio tracks into stems such as vocals, drums, bass, and other instruments. Spleeter offers separation capabilities including 2, 4, and 5 stem models. The architecture leverages deep learning models trained on large datasets, enabling it to decompose audio files with impressive speed, reaching 100x real-time on GPUs for 4-stem separation. It can be used via command-line or integrated as a Python library. Use cases include music production, remixing, audio restoration, and AI-driven music analysis. The tool supports pip and Docker installations.