Polymath

Polymath converts your music library into a searchable, quantized production toolkit. Using a pipeline of open-source machine learning models, it isolates individual sound elements from full mixes, detects and normalizes tempo across tracks, identifies structural and tonal characteristics, and transcribes audio directly into MIDI for creative manipulation and reuse.

Highlights

• Stem separation via Demucs neural networks isolates drums, bass, vocals, and other instruments with precision
• Audio-to-MIDI conversion captures melodies and harmonies in editable format for rearrangement
• Automatic tempo detection and time-stretching aligns sample libraries to a unified beat grid
• Musical structure analysis reveals sections, keys, and tonal content automatically
• Free and open-source with Docker support for reproducible, serverless processing pipelines

Whether you're organizing sample packs, creating mashups, or exploring generative music composition, Polymath eliminates tedious manual audio preparation. Use it as a Python library for integration or via the command-line interface. Fully free under MIT licensing.