+26
Daily
8,970
stars
1,125
forks
133
watchers
807
prs
1,316
issues
49
releases
README
librosa
librosa is a Python library for audio and music signal processing. It provides the foundational algorithms and tools required for building music information retrieval (MIR) systems.
Documentation
Full documentation is available at:
https://librosa.org/doc/
If you're new to librosa, we recommend starting with the
tutorials.
If you're looking for API details, head directly to the API reference.
To see what's new, check out the change log.
Installation
Using PyPI
The latest stable release is available on PyPI, and you can install it by the command
pip install librosa
Using conda
Anaconda users can install using conda-forge:
conda install -c conda-forge librosa
For optional dependencies and advanced setup, see the guide in the documentation.
---
Citing
There are two ways to cite librosa in scholarly work, depending on whether you are citing it for its use in your work
or for its design and motivation.
We additionally encourage you to cite the original publications describing the methods implemented in librosa, in addition to citing librosa itself. References can often be found within function documentation.
Citing for use
If you are citing for use in your own work, please cite the version you used by retrieving the appropriate DOI and citation information from Zenodo.
This ensures that all contributing authors are credited for the work you are building on.
From librosa version 0.10.2 or later, you can also use librosa.cite() to get the DOI link for any version of librosa:
python
import librosa
librosa.cite()
And from 1.0 onward, you can directly retrieve the BibTeX entry for the version you are using:
python
print(librosa.cite(bib=True))
Citing for design
If you wish to cite librosa for its design, motivation, etc., please cite the paper
published at SciPy 2015:
McFee, Brian, Colin Raffel, Dawen Liang, Daniel PW Ellis, Matt McVicar, Eric Battenberg, and Oriol Nieto.
"librosa: Audio and music signal analysis in python."
In Proceedings of the 14th python in science conference, pp. 18-25. 2015.
As a rule of thumb:
- Use a versioned librosa citation to document the software implementation used in your experiments.
- Cite the original methodological papers for the algorithms that are central to your work.
- Cite the SciPy paper when discussing librosa itself, its design, or its contribution as a software library.Top issues
#1843Librosa on Debian!25 comments · 2024-05-30#1364Support user-supplied time format strings in display20 comments · 2021-08-17#1279Consider (de)modulating stft18 comments · 2021-02-04#18Make filterbanks generic16 comments · 2013-07-31#2067[ENH]: Extend precise return type annotations to the feature and onset modules14 comments · 2026-07-07#1806Add tuning param to hz_to_note()12 comments · 2024-02-14#1278stft: time dimension might be off12 comments · 2021-02-04#1503Frequency range data10 comments · 2022-06-11
Latest release
1.0.1 1.0.1 2026-11-12What's Changed Keep doctrees and reST sources out of the published docs by @Kayvan-Zahiri in https://github.com/librosa/librosa/pull/2089 making linkcheck failures not catastrophic for CI by @bmcfee in https://github.com/librosa/librosa/pull/2093 Revising linkcheck action again by @bmcfee in https://github.com/librosa/librosa/pull/2098 Forward window to stft and istft in effects.hpss, harmonic and percussive by @isaacsight in https://github.com/librosa/librosa/pull/2097 Enable Ruff ARG unused-argument checks and resolve findings by @rajit-sareen in https://github.com/librosa/librosa/