Skip to content

Unofficial implementation of MT3: Multi-Task Multitrack Music Transcription (Google Research, 2022) in pytorch

Notifications You must be signed in to change notification settings

rlax59us/MT3-pytorch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MT3-pytorch for MAESTRO dataset

Now, this is an unofficial implementation of MT3 for single track(SEQUENCE-TO-SEQUENCE PIANO TRANSCRIPTION WITH TRANSFORMERS) in pytorch.

Converted original model code in MT3 repository from jax to pytorch.

Later, I will update to extend the model to multi-track and multi-task for implementing the MT3 model with Slakh2100 dataset.

Prerequisite

First of all, please install the appropriate version of pytorch library.

With Anaconda, you can install using below command line(example).

$ conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=11.0 -c pytorch 

After then, install the specified libraries in requirements.txt file.

$ pip install -r requirements.txt

Usage

Not done yet.

Train

$ python train.py

Results

Citations

@article{
  title={SEQUENCE-TO-SEQUENCE PIANO TRANSCRIPTION WITH TRANSFORMERS},
  author={Curtis Hawthorne, Ian Simon, Rigel Swavely, Ethan Manilow and Jesse Engel},
  paper={https://arxiv.org/abs/2107.09142v1},
  year={2021}
}

About

Unofficial implementation of MT3: Multi-Task Multitrack Music Transcription (Google Research, 2022) in pytorch

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages