Open-Unmix Paper

This repository combines the software contributions for open-unmix, a reference implementation for deep learning based music source separation.

We choose PyTorch to serve as a reference implementation for this submission due to its balance between simplicity and modularity. Furthermore, we already ported the core model to NNabla and plan to release a port for Tensorflow 2.0, once the framework is released. Note that the ports will not include pre-trained models as we cannot make sure the ports would yield identical results, thus leaving a single baseline model for researchers to compare with

Software Packages

Open-Unmix for Pytorch

• Code: open-unmix-pytorch
• Status: feature complete
• Tag: v1.0.0
• Pretrained models: UMXHQ and UMX
• 

musdb dataset parser

A python package to parse and process the MUSDB18 dataset, the largest open access dataset for music source separation.

• Code: musdb
• Tag: v0.3.1
• Status: released on pypi in version 0.3.1
• 

museval objective evaluation

• Code: museval
• Tag: v0.3.0
• Status: released on pypi in version 0.3.0
• 

norbert: wiener filter implementations

• Code: norbert
• Status: released on pypi in version 0.2.1
• Tag: v0.2.1
• 

Paper

to create the paper locally

docker run -v $PWD:/data openbases/openbases-pdf pdf