OpenSoundscape is a utility library for analyzing bioacoustic data. It consists of command line scripts for tasks such as preprocessing audio data, training machine learning models to classify vocalizations, estimating the spatial location of sounds, identifying which species’ sounds are present in acoustic data, and more.
These utilities can be strung together to create data analysis pipelines. OpenSoundscape is designed to be run on any scale of computer: laptop, desktop, or computing cluster.
OpenSoundscape is currently in active development. If you find a bug, please submit an issue. If you have another question about OpenSoundscape, please email Sam Lapp (
pitt.edu) or Tessa Rhinehart (
For examples of some of the utilities offered, please see the “Tutorials” section of the documentation. Included are instructions on how to download and use a pretrained machine learning model from our publicly available set of models. We plan to add additional tutorials soon.
OpenSoundscape can be installed either via pip (for users) or poetry (for developers contributing to the code). Either way, Python 3.7 or higher is required.
Installation via pip (most users)¶
Just give me the pip command!¶
Already familiar with installing python packages via pip? The pip command to install OpenSoundscape is
Python 3.7 is required to run OpenSoundscape. Download it from this website.
We recommend installing OpenSoundscape in a virtual environment to prevent dependency conflicts. Below are instructions for installation with Python’s included virtual environment manager,
venv, but feel free to use another virtual environment manager (e.g.
virtualenvwrapper) if desired.
Run the following commands in your bash terminal:
- Check that you have installed Python 3.7._:
- Change directories to where you wish to store the environment:
cd [path for environments folder]
- Tip: You can use this folder to store virtual environments for other projects as well, so put it somewhere that makes sense for you, e.g. in your home directory.
- Make a directory for virtual environments and
mkdir .venv && cd .venv
- Create an environment called
opensoundscapein the directory:
python3 -m venv opensoundscape
- For Windows computers: activate/use the environment:
- For Mac computers: activate/use the environment
- Install OpenSoundscape in the environment:
pip install opensoundscape==0.4.1
- Once you are done with OpenSoundscape, deactivate the environment:
- To use the environment again, you will have to refer to absolute path of the virtual environments folder. For instance, if I were on a Mac and created
.venvinside a directory
/Users/MyFiles/CodeI would activate the virtual environment using:
For some of our functions, you will need a version of
ffmpeg >= 0.4.1. On Mac machines,
ffmpeg can be installed via
Installation via poetry (contributors and advanced users)¶
Poetry installation allows direct use of the most recent version of the code. This workflow allows advanced users to use the newest features in OpenSoundscape, and allows developers/contributors to build and test their contributions.
To install via poetry, do the following:
Figure out where the virtualenvwrapper.sh file is:
Add the following to your
~/.bashrcand source it. .. code-block:
# virtualenvwrapper + poetry export PATH=~/.local/bin:$PATH export WORKON_HOME=~/Library/Caches/pypoetry/virtualenvs source [insert path to virtualenvwrapper.sh, e.g. ~/.local/bin/virtualenvwrapper_lazy.sh]
Users: clone this github repository to your machine:
git clone https://github.com/kitzeslab/opensoundscape.git
Contributors: fork this github repository and clone the fork to your machine
Ensure you are in the top-level directory of the clone
Switch to the development branch of OpenSoundscape:
git checkout develop
Build the virtual environment for opensoundscape:
- If poetry install outputs the following error, make sure to download Python 3.7:
- If you are on a Mac and poetry install fails to install
numba, contact one of the developers for help troubleshooting your issues.
Activate the virtual environment with the name provided at install e.g.:
Check that OpenSoundscape runs:
Run tests (from the top-level directory):
poetry run pytest
Go back to your system’s Python when you are done:
To use OpenSoundscape within JupyterLab, you will have to make an
for the OpenSoundscape virtual environment.
- Activate poetry virtual environment, e.g.:
poetry env listif you’re not sure what the name of the environment is
- Create ipython kernel:
python -m ipykernel install --user --name=[name of poetry environment] --display-name=OpenSoundscape
- Now when you make a new document on JupyterLab, you should see a Python kernel available called OpenSoundscape.
- Contributors: if you include Jupyter’s
autoreload, any changes you make to the source code installed via poetry will be reflected whenever you run the
%autoreloadline magic in a cell:
Contributing to code¶
Make contributions by editing the code in your fork. Create branches
for features using
git checkout -b feature_branch_name and push these
changes to remote using
git push -u origin feature_branch_name. To merge a
feature branch into the development branch, use the GitHub
web interface to create a merge request.
When contributions in your fork are complete, open a pull request using the GitHub web interface. Before opening a PR, do the following to ensure the code is consistent with the rest of the package:
- Run tests:
poetry run pytest
- Format the code with
blackstyle (from the top level of the repo):
- Additional libraries to be installed should be installed with
poetry add, but in most cases contributors should not add libraries.
Contributing to documentation¶
Build the documentation using either poetry or sphinx-build
- With poetry:
poetry run build_docs
- With sphinx-build:
sphinx-build doc doc/_build
Publish the documentation with the following commands: