spotify/basic-pitch

Employs a lightweight instrument-agnostic neural network that supports polyphonic transcription with multipitch estimation, trained by Spotify's Audio Intelligence Lab and published at ICASSP 2022. Provides multiple model serialization formats (TensorFlow, CoreML, TensorFlowLite, ONNX) with platform-specific defaults to minimize dependencies—TensorFlow on Python 3.11+, CoreML on macOS, TensorFlowLite on Linux, and ONNX on Windows. Exposes both CLI and programmatic Python APIs with configurable frequency filtering and batch processing support via preloaded model objects.

4,764 stars.