This directory contains Magenta models.
- Arbitrary Image Stylization: A machine learning system for performing fast artistic style transfer that may work on arbitrary painting styles.
- Coconet: Counterpoint by Convolution train a convolutional neural network to complete partial musical scores.
- Drums RNN: Applies language modeling to drum track generation using an LSTM.
- GANSynth: GANSynth is an algorithm for synthesizing audio with generative adversarial networks.
- Image Stylization: A "Multistyle Pastiche Generator" that generates artistics representations of photographs. Described in A Learned Representation For Artistic Style.
- Improv RNN: Generates melodies a la Melody RNN, but conditions the melodies on an underlying chord progression.
- Melody RNN: Applies language modeling to melody generation using an LSTM.
- Music VAE: A hierarchical recurrent variational autoencoder for music.
- NSynth: "Neural Audio Synthesis" as described in NSynth: Neural Audio Synthesis with WaveNet Autoencoders.
- Onsets and Frames: Automatic piano music transcription model as described in Onsets and Frames: Dual-Objective Piano Transcription
- Performance RNN: Applies language modeling to polyphonic music using a combination of note on/off, timeshift, and velocity change events.
- Piano Genie: Piano Genie is a system for learning a low-dimensional discrete representation of piano music. It uses an encoder RNN to compress piano sequences (88 keys) into many fewer buttons (e.g. 8). A decoder RNN is responsible for converting the simpler sequences back to piano space.
- Pianoroll RNN-NADE: Applies language modeling to polyphonic music generation using an LSTM combined with a NADE, an architecture called an RNN-NADE. Based on the architecture described in Modeling Temporal Dependencies in High-Dimensional Sequences: Application to Polyphonic Music Generation and Transcription.
- Polyphony RNN: Applies language modeling to polyphonic music generation using an LSTM. Based on the BachBot architecture described in Automatic Stylistic Composition of Bach Choralies with Deep LSTM.
- RL Tuner: Takes an LSTM that has been trained to predict the next note in a monophonic melody and enhances it using reinforcement learning (RL). Described in Tuning Recurrent Neural Networks with Reinforcement Learning and Sequence Tutor: Conservative Fine-Tuning of Sequence Generation Models with KL-control
- Score2Perf and Music Transformer: Score2Perf is a collection of Tensor2Tensor problems for generating musical performances, either unconditioned or conditioned on a musical score.
- Sketch RNN: A recurrent neural network model for generating sketches. Described in Teaching Machines to Draw and A Neural Representation of Sketch Drawings.
- SVG VAE: SVG VAE is a Tensor2Tensor problem for generating font SVGs. Described in A Learned Representation for Scalable Vector Graphics.