Generating Music with a Self-Correcting Non-Chronological Autoregressive Model
arXiv:2008.08927
Abstract
We describe a novel approach for generating music using a self-correcting, non-chronological, autoregressive model. We represent music as a sequence of edit events, each of which denotes either the addition or removal of a note---even a note previously generated by the model. During inference, we generate one edit event at a time using direct ancestral sampling. Our approach allows the model to fix previous mistakes such as incorrectly sampled notes and prevent accumulation of errors which autoregressive models are prone to have. Another benefit is a finer, note-by-note control during human and AI collaborative composition. We show through quantitative metrics and human survey evaluation that our approach generates better results than orderless NADE and Gibbs sampling approaches.
arXiv comment8 pages, 4 figures
References in corpus (12)
- Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
- Generating Long Sequences with Sparse Transformers
- C-RNN-GAN: Continuous recurrent neural networks with adversarial training
- Professor Forcing: A New Algorithm for Training Recurrent Networks
- MidiNet: A Convolutional Generative Adversarial Network for Symbolic-domain Music Generation
- How (not) to Train your Generative Model: Scheduled Sampling, Likelihood, Adversary?
- Insertion Transformer: Flexible Sequence Generation via Insertion Operations
- MuseGAN: Multi-track Sequential Generative Adversarial Networks for Symbolic Music Generation and Accompaniment
- A Deep and Tractable Density Estimator
- Counterpoint by Convolution
- KERMIT: Generative Insertion-Based Modeling for Sequences
- The Bach Doodle: Approachable music composition with machine learning at scale