Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/yoyololicon/danna-sep
https://github.com/yoyololicon/danna-sep
Last synced: 21 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/yoyololicon/danna-sep
- Owner: yoyololicon
- Created: 2021-08-27T15:29:25.000Z (about 3 years ago)
- Default Branch: master
- Last Pushed: 2022-03-12T15:47:01.000Z (over 2 years ago)
- Last Synced: 2024-10-04T16:24:57.565Z (about 1 month ago)
- Language: Python
- Size: 12.7 KB
- Stars: 12
- Watchers: 2
- Forks: 3
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Danna-Sep: Danna Sama no Source Separation
[](https://huggingface.co/spaces/yoyololicon/Danna-Sep)
The winning model I used in [MDX 2021 Challenge](https://www.aicrowd.com/challenges/music-demixing-challenge-ismir-2021/winners).
## Installation
```commandline
python setup.py install
```
## Usage
```
usage: danna_sep [-h] [--outdir OUTDIR] [--fast] infile
positional arguments:
infile input audio file
optional arguments:
-h, --help show this help message and exit
--outdir OUTDIR output directory. Default to current working directory
--fast faster inference using only two of the models
```
Given an audio file, the program will split it into 4 stems, which are drums, bass, vocals and other, and store them in the given directory as `.wav` files.
When execute the first time it will download our pre-trained models (around 1 to 2 Gb) to the directory specified by the environment variable `DANNA_CHECKPOINTS`, which by default is `~/danna-sep-checkpoints`.
This process is very time comsuming and require at least 16 Gb of RAM.
## Training
Please refer to our [training repo](https://github.com/yoyololicon/music-demixing-challenge-ismir-2021-entry).
## Web Demo
Try Danna-Sep on [Huggingface Spaces](https://huggingface.co/spaces/yoyololicon/Danna-Sep).
## TODO
- [ ] convert to ONNX format
- [ ] Pack it as standalone app