A collection of different Model Architectures for Music Generation. These include LSTMs, GRUs, BiGRUs, Attention based Models and Transformers for genenration of Music in its Symbolic Representation, i.e. MIDI Based representation.
This repo is a case study of various Algorithms for music Generation, where the models are present as .ipynb Notebooks which can run in Google Colab. Certain example results from the training has also been uploaded to the repository.
- Simple LSTM
- 1D Convolution
- LSTM + DropOut
- GRU + Attention
- BiGRU + Attention
- Transformer Baseline
- Encoder Only
- Encoder + Decoder - NMT style generation
- Music Transformer - with Relative Attention
- Encoder Only
- Encoder + Decoder
- Compound Word Transformer or CP - Transformer ( Colab Notebook )