https://github.com/fakufaku/fast_bss_eval
A fast implementation of bss_eval metrics for blind source separation
https://github.com/fakufaku/fast_bss_eval
audio-processing blind-source-separation bss evaluation-metrics fast
Last synced: 2 months ago
JSON representation
A fast implementation of bss_eval metrics for blind source separation
- Host: GitHub
- URL: https://github.com/fakufaku/fast_bss_eval
- Owner: fakufaku
- License: mit
- Created: 2021-10-12T08:11:25.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2022-06-07T01:37:15.000Z (about 3 years ago)
- Last Synced: 2024-06-11T19:59:07.987Z (about 1 year ago)
- Topics: audio-processing, blind-source-separation, bss, evaluation-metrics, fast
- Language: Python
- Homepage: https://fast-bss-eval.readthedocs.io/en/latest/
- Size: 1.11 MB
- Stars: 126
- Watchers: 4
- Forks: 8
- Open Issues: 4
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.rst
- License: LICENSE
Awesome Lists containing this project
README
fast\_bss\_eval
==============
[](https://badge.fury.io/py/fast-bss-eval)
[](https://fast-bss-eval.readthedocs.io/en/latest/?badge=latest)
[](https://github.com/fakufaku/fast_bss_eval/actions/workflows/lint.yml)
[](https://github.com/fakufaku/fast_bss_eval/actions/workflows/pythonpackage.yml)
[](https://codecov.io/gh/fakufaku/fast_bss_eval)
> Do you have a zillion BSS audio files to process and it is taking days ?
> Is your simulation never ending ?
>
> Fear no more! `fast_bss_eval` is here to help **you!**
`fast_bss_eval` is a fast implementation of the bss\_eval metrics for the
evaluation of blind source separation. Our implementation of the bss\_eval
metrics has the following advantages compared to other existing ones.
* seamlessly works with **both** [numpy](https://numpy.org/) arrays and [pytorch](https://pytorch.org) tensors
* very fast
* can be even faster by using an iterative solver (add `use_cg_iter=10` option to the function call)
* differentiable via pytorch
* can run on GPU via pytorch
Author
------
* [Robin Scheibler]([email protected])
Quick Start
-----------
### Install
```bash
# from pypi
pip install fast-bss-eval
# or from source
git clone https://github.com/fakufaku/fast_bss_eval
cd fast_bss_eval
pip install -e .
```
### Use
Assuming you have multichannel signals for the estmated and reference sources
stored in wav format files names ``my_estimate_file.wav`` and
``my_reference_file.wav``, respectively, you can quickly evaluate the bss_eval
metrics as follows.
```python
from scipy.io import wavfile
import fast_bss_eval
# open the files, we assume the sampling rate is known
# to be the same
fs, ref = wavfile.read("my_reference_file.wav")
_, est = wavfile.read("my_estimate_file.wav")
# compute the metrics
sdr, sir, sar, perm = fast_bss_eval.bss_eval_sources(ref.T, est.T)
```
Benchmark
---------
This package is significantly faster than other packages that also allow
to compute bss\_eval metrics such as [mir\_eval](https://github.com/craffel/mir_eval) or [sigsep/bsseval](https://github.com/sigsep/bsseval).
We did a benchmark using numpy/torch, single/double precision floating point
arithmetic (fp32/fp64), and using either Gaussian elimination or a [conjugate
gradient descent](https://en.wikipedia.org/wiki/Conjugate_gradient_method)
(solve/CGD10).
Citation
--------
If you use this package in your own research, please cite [our paper](https://arxiv.org/abs/2110.06440) describing it.
```bibtex
@misc{scheibler_sdr_2021,
title={SDR --- Medium Rare with Fast Computations},
author={Robin Scheibler},
year={2021},
eprint={2110.06440},
archivePrefix={arXiv},
primaryClass={eess.AS}
}
```
License
-------
2021 (c) Robin Scheibler, LINE Corporation
All of this code is released under [MIT License](https://opensource.org/licenses/MIT) with the exception of `fast_bss_eval/torch/hungarian.py` which is under [3-clause BSD License](https://opensource.org/licenses/BSD-3-Clause).