Whombat

Whombat is an open-source, web-based audio annotation tool designed to streamline audio data labeling and annotation, with a particular focus on supporting machine learning model development.

Installation

Visit the Releases section on GitHub to download the bundled version compatible with your operating system.

If you prefer installing Whombat via Python, run the command

pip install whombat

For detailed installation instructions, refer to the Installation section of the documentation.

Usage

To run whombat either click on the bundled executable or run

python -m whombat

We have prepared a User Guide to accompany you in your annotation work. There you will be able to see all the features provided by Whombat, as well as clear instructions on how to use them.

Contribution

As a open source project we are incredibly excited for having contributions from the community. Head over to the Contributions section of the documentation to see how you can contribute.

Development

This repository contains two apps:

• back/: the Python backend, managed with uv
• front/: the Next.js frontend, managed with npm

Common development tasks are available from the repository root with just.

Install all development dependencies:

just install

Run the backend and frontend in development mode:

just dev

Run the main checks for both apps:

just check

Useful commands:

just dev-back
just dev-front
just docs
just storybook

Citation

If you want to use Whombat for your research, please cite as:

Balvanera, S. M., Mac Aodha, O., Weldy, M. J., Pringle, H., Browning, E., & Jones, K. E. (2023). Whombat: An open-source annotation tool for machine learning development in bioacoustics. arXiv preprint arXiv:2308.12688.

Acknowledgements

Whombat has been developed with the generous support of the Mexican Council of the Humanities, Science and Technology (CONAHCyT; Award Number 2020-000017-02EXTF-00334) and University College London (UCL).