EDANSA: Pre-trained Bioacoustic Models

This repository provides pre-trained models for detecting bioacoustic events and specific bird species in Passive Acoustic Monitoring (PAM) data, developed as part of the EDANSA project.

Available Models

Two primary pre-trained models are available in the assets/ directory:

1. General Bioacoustics Model (31m2plxv-V1): Detects broader sound categories (Biophony, Anthropophony, Birds, Insects, Silence, etc.). See Model Card.
2. Bird Species Model (ppq7zxqq): Detects specific North American arctic bird species (Lapland Longspur, White-crowned Sparrow, etc.). See Model Card.

Installation

1. Clone the repository:

   git clone https://github.com/speechLabBcCuny/EDANSA.git
   cd EDANSA

2. Create Conda Environment (Recommended): Choose the environment file based on your system (GPU or CPU).

   • For GPU-enabled systems (e.g., HPC nodes with NVIDIA GPUs):

     micromamba create -f environment_gpu.yml -n <your_env_name>
     micromamba activate <your_env_name>

     (This environment uses pytorch-gpu and CUDA-enabled torchaudio.)
   • For CPU-only systems (e.g., local machines, CPU-only servers):

     micromamba create -f environment_cpu.yml -n <your_env_name>
     micromamba activate <your_env_name>

     (This environment uses CPU-only pytorch and torchaudio.)

   (See Full Installation Guide for details, alternative Conda/Mamba commands, Pip instructions, and troubleshooting.)

Basic Inference Usage

Run inference using the runs/augment/inference.py script. Here's an example using the general model:

python runs/augment/inference.py \
    --model_path assets/31m2plxv-V1/model_info/best_model_370_val_f1_min=0.8028.pt \
    --config_file assets/31m2plxv-V1/model_info/model_config.json \
    --input_folder /path/to/your/audio/ \
    --output_folder /path/to/save/results/

(See Running Inference Guide for more details and arguments.)

Documentation

For complete details on installation, usage, data preparation, model performance, and troubleshooting, please refer to the Full Documentation Site (built from the docs/ directory).

Citation

If you use the code or models from this repository, please cite:

@inproceedings{Coban2022,
    author = "\c{C}oban, Enis Berk and Perra, Megan and Pir, Dara and Mandel, Michael I.",
    title = "EDANSA-2019: The Ecoacoustic Dataset from Arctic North Slope Alaska",
    booktitle = "Proceedings of the 7th Detection and Classification of Acoustic Scenes and Events 2022 Workshop (DCASE2022)",
    address = "Nancy, France",
    month = "November",
    year = "2022",
    abstract = "The arctic is warming at three times the rate of the global average, affecting the habitat and lifecycles of migratory species that reproduce there, like birds and caribou. Ecoacoustic monitoring can help efficiently track changes in animal phenology and behavior over large areas so that the impacts of climate change on these species can be better understood and potentially mitigated. We introduce here the Ecoacoustic Dataset from Arctic North Slope Alaska (EDANSA-2019), a dataset collected by a network of 100 autonomous recording units covering an area of 9000 square miles over the course of the 2019 summer season on the North Slope of Alaska and neighboring regions. We labeled over 27 hours of this dataset according to 28 tags with enough instances of 9 important environmental classes to train baseline convolutional recognizers. We are releasing this dataset and the corresponding baseline to the community to accelerate the recognition of these sounds and facilitate automated analyses of large-scale ecoacoustic databases."
}

License

This project is licensed under the MIT License - see the LICENSE.md file for details.