gaborfodor / wave-soundscape-annotation

Setup

Clone the repo and create python environment with the required dependencies.

python -m venv venv
source venv/bin/activate
python -m pip install --upgrade pip
pip install -r requirements.txt
# You might need ffmpeg/libsndfile or other system dependencies to get librosa working

Download & extract Wave

Optional Download all the required audio chunks and pretrained model from kaggle.

kaggle datasets download gaborfodor/rainforestexampleaudiochunks
mv rainforestexampleaudiochunks.zip data/
unzip data/rainforestexampleaudiochunks.zip

Run the app

1. Start the Wave development server: $ ./waved
2. Start the app: $ wave run rainforest.dashboard
3. Go to http://localhost:10101/

Annotate

Select any species and fix the collected labels.

Recognize

Upload your own rainforest recording and analyze it with a single pretrained CNN14 model.