What is Polymath?
Polymath is a cutting-edge machine learning tool designed to transform any music library—whether stored on a hard drive or sourced from YouTube—into a fully searchable, production-ready sample library. By leveraging advanced audio processing and ML models, Polymath automates the complex tasks of stem separation, tempo quantization, musical structure analysis, key detection, and audio-to-MIDI conversion. This allows music producers, DJs, and ML audio developers to effortlessly extract and repurpose individual elements from any song, such as beats, basslines, melodies, and vocals, all synchronized to a consistent tempo and beat grid (e.g., 120 BPM).
Key features include automatic separation of songs into stems (drums, bass, vocals, etc.), quantization to a user-defined tempo, analysis of musical sections (verse, chorus, bridge), key and chord detection, timbre and loudness profiling, and conversion of audio to MIDI for further editing. The resulting sample library is fully searchable, enabling users to quickly find and combine elements from different tracks to create unique compositions. For example, a producer can grab a drum loop from a Funkadelic track, a bassline from Tito Puente, and horn stabs from Fela Kuti, all seamlessly integrated into their DAW. DJs can use Polymath to create polished, hour-long mash-up sets by discovering related tracks and mixing them with perfect harmonic and rhythmic alignment.
For ML developers, Polymath simplifies dataset creation for training generative models by extracting and labeling audio features at scale. The tool supports GPU acceleration via CUDA for faster processing. Installation requires Python 3.7–3.10 and dependencies like TensorFlow and PyTorch. A Dockerfile is also provided for containerized deployment. Polymath is open-source and available on GitHub, with community support through Discord. Whether you're a producer seeking creative inspiration, a DJ crafting seamless mixes, or a researcher building audio datasets, Polymath streamlines your workflow and unlocks new possibilities in music production and analysis.
Use Cases
Key Features
stem separation tempo quantization musical structure analysis key detection audio to MIDI conversion timbre profiling loudness profiling searchable sample libraryWho is it for?
music producers, DJs, ML audio developers, sound designers, composers, remix artists
