Magenta.js is a collection of TypeScript libraries for doing inference with pre-trained Magenta models. All libraries are published as npm packages.
Learn more about the Magenta project at our blog and main repo.
- core: A shared library containing core functions for processing NoteSequence protobufs, the primary data format in Magenta's music models. Implements a subset of the Python "Magenta Music" library.
- protobuf: A shared library containing the TypeScript implementation of NoteSequence protobufs.
- music_rnn: A model library containing a deeplearn.js implementation of Magenta's RNN-based music generation models including MelodyRNN, DrumsRNN, PerformanceRNN, and PolyphonyRNN, whose Python TensorFlow implementations can be found in the main Magenta repo.
- music_vae: A model library containing MusicVAE.js, a deeplearn.js implementation of Magenta's MusicVAE model. The Python TensorFlow implementation can be found in the main Magenta repo.