Welcome to the public model repository of the The HALLO (Humans and ALgorithms Listening for Orcas) project.

Here, you will find the instructions, scripts and configuration files required to train deep learning models at detecting and classifying vocalisations made by killer whales.

Of course, instructions and configuration files alone won't do. You will also need access to large bodies of annotated underwater sound recordings. Unfortunately, these recordings are not yet publicly available, although we are working feverishly to change this. Please check in here again soon for instructions on how to access the data.

Currently, the only available model is a binary classifier for detecting killer whale vocalisations (pulsed calls, mainly). To find out more about this model, click the link below.

• kw-detector

More models, including models capable of classifiying vocalisations at finer levels of resolution, e.g., according to ecotype or even pod, will be released in coming years.

If you have questions or comments about this repository, please submit an issue. For general inquiries about HALLO, please contact the project's PI, Ruth Joy.

If you wish to become part of HALLO, please contact the project's PI, Ruth Joy.

HALLO is funded by Fisheries and Oceans Canada (DFO) through the EOSCF (2019-2021) and CNFASAR (2022-2026) programs.

The project is licensed under the GNU GPLv3 license and hence the contents of this repository are freely available for anyone to use and modify.

As of June 2022, this project is under active development.